Process planning for precision manufacturing An approach based on methodological studies Mats Bagge
Doctoral Thesis 2014 KTH Royal Institute of Technology Engineering Sciences Department of Production Engineering SE-100 44 Stockholm, Sweden
TRITA IIP-14-04 ISSN 1650-1888 ISBN 978-91-7595-172-0 Akademisk avhandling som med tillstånd av KTH i Stockholm framlägges till offentlig granskning för avläggande av teknisk doktorsexamen tisdagen den 10 juni 2014 kl. 13:00 i sal M311, KTH Industriell Produktion, Brinellvägen 68, Stockholm Copyright © Mats Bagge 2014 Tryck: Universitetsservice US AB
III
Abstract Process planning is a task comprising a broad range of activities to design and develop an appropriate manufacturing process for producing a part. Interpretation of the part design, selection of manufacturing processes, definition of operations, operation sequences, machining datums, geometrical dimensions and tolerances are some common activities associated with the task. Process planning is also “the link between product design and manufacturing” with the supplementary commission to support design of competitive products. Process planning is of a complex and dynamic nature, often managed by a skilled person with few, or no, explicit methods to solve the task. The work is heuristic and the result is depending on personal experiences and decisions. Since decades, there have been plenty of attempts to develop systems for computer-aided process planning (CAPP). CAPP is still awaiting its breakthrough and one reason is the gap between the functionality of the CAPP systems and the industrial process planning practice. This thesis has an all-embracing aim of finding methods that cover essential activities for process planning, including abilities to predict the outcome of a proposed manufacturing process. This is realised by gathering supporting methods suitable to manage both qualitative and quantitative characterisation and analyses of a manufacturing process. The production research community has requested systematisation and deeper understanding of industrial process planning. This thesis contributes with a flow chart describing the process planning process (PPP), in consequence of the methodological studies. The flow chart includes process planning activities and information flows between these activities. The research has been performed in an industrial environment for high volume manufacturing of gear parts. Though gear manufacturing has many distinctive features, the methods and results presented in this thesis are generally applicable to precision manufacturing of many kinds of mechanical parts.
Keywords Process planning, precision manufacturing, machining, tolerance chain analysis, process behaviour, process performance, process capability, in-process workpiece.
V
Acknowledgements This thesis is a result of my work as an industrial PhD student at the Department of Production Engineering IIP, KTH Royal Institute of Technology, employed by Scania CV AB, Södertälje. Since the start in 2006, I have been involved in several research projects where collaborations between Swedish industry, academia and research institutes have been a common denominator. These projects have provided many interesting research topics related to my main field of research – process planning – and given many fruitful opportunities to get to know nice and proficient people. Among these people, I would first like to express my special appreciation and thanks to my supervisors Professor Bengt Lindberg and Professor Sören Andersson at KTH and Ulf Bjarre at Scania CV AB for inspiring cooperation over the years of studies. The appended publications in this thesis have also been written in an invigorating manner together with Professor Cornel Mihai Nicolescu, Mats Werke and Mikael Hedlind. Thank you! I would also like to thank my colleague and participant in the KUGG project, Julia Lundberg Gerth for taking time and giving well advises when writing this thesis. My appreciations to the other Ph.D. students in the KUGG project (where it all begun); Mathias Werner, Ellen Bergseth, Sören Sjöberg and Karin Björkeborn. One important source for inspiration and discussions about crossdisciplinary considerations related to production research has been participating in the seminars lead by Associate Professor Peter Gröndahl. Many persons have contributed to the engaging enthusiasm in the seminar group, especially Robert Gerth, Tord Johansson, Joakim Storck, Jens von Axelson, Richard Lindqvist and Magnus Lundgren. I have in addition had the favour to get to know and spend time together with many people at IIP, in front of others; Anders Berglund, Andreas Archenti and Lorenzo Daghini. It has been a pleasure to cadge your offices! - Marie, Johan, Maj-Britt and Mona, let me express my sincere gratitude for all your assistance when colloquial duties have not fit into my business schedule! Finally, all my love goes to Carolin, Hanna and Ellen. You are great! Mats Bagge, Södertälje, May 2014
VII
List of publications This thesis is based on five publications of which one is a licentiate thesis and four are papers. The four papers are appended at the end of this doctoral thesis. Licentiate thesis (not appended) Bagge, M., 2009. An approach for systematic process planning of gear transmission parts. Licentiate thesis, Stockholm, Sweden: Royal Institute of Technology. Paper A Werke, M., Bagge, M., Nicolescu, M. & Lindberg, B., 2014. Process modelling using upstream analysis of manufacturing sequences. Submitted to: The International Journal of Advanced Manufacturing Technology (Under review April 2014). Paper B Bagge, M. & Lindberg, B., 2012. Analysis of process parameters during press quenching of bevel gear parts. In: M. Björkman, ed. Proceedings of The 5th International Swedish Production Symposium. Linköping: Produktionsakademien, pp. 251-259. Paper C Bagge, M., Hedlind, M. & Lindberg, B., 2013. Tolerance chain design and analysis of in-process workpiece. In: A. Archenti & A. Maffei, eds. Proceedings of the International Conference on Advanced Manufacturing Engineering and Technologies. Stockholm: KTH Royal Institute of Technology, pp. 305-315. Paper D Bagge, M., Hedlind, M. & Lindberg, B., 2014. Process chain based workpiece variation simulation for performance utilisation analysis. Submitted to: Swedish Production Symposium 2014, Göteborg, Sweden.
IX
Abbreviations AI: AIAG: APQP: CAD: CAM: CAPP: CoPQ: DDC: DDC2: DOE: DPMO: FD: FFI: Icam: IDEF: IPW: KPI: NC: OE: OEM: PCI: PLM: PPAP: P-FMEA: PPM: PPP: PUR: R&D: RE: ReVer: SE: T1, T2... : TE: VW:
Artificial Intelligence Automotive Industry Action Group Advanced Product Quality Computer Aided Design Computer Aided Manufacturing Computer Aided Process Planning Cost of Poor Quality Dimension Dependency Chart Dimension Dependency Chart 2 Design of Experiments Defects per Million Opportunities Final Dimension Fordonsstrategisk Forskning och Innovation Integrated Computer Aided Manufacturing Integration DEFfinition In Process Workpiece Key Performance Index Numerical Control Operation Element Original Equipment Manufacturer Process Capability Index Product Lifecycle Management Production Part Approval Process Process Failure Mode and Effects Analysis Parts Per Million Process Planning Process Performance Utilisation Ratio Research and Development Random Error Realistic Verification Systematic Error Tool number 1, 2 et cetera Total Error Virtual Workshop
XI
Contents List of publications...................................................................................VII Abbreviations............................................................................................IX
1 Introduction ..................................................................... 1 1.1 Process planning for manufacturing ............................................... 1 1.2 Structure of the thesis ..................................................................... 4
2 Frame of reference .......................................................... 5 2.1 PPP - the process planning process ............................................... 5 2.2 Process capability – a boundary condition for process planning .... 7 2.3 Conceptions of process characterisation ........................................ 8 2.4 The beam balance .......................................................................... 8 2.5 The beam balance versus process capability ................................. 9 2.6 Transmission part manufacturing .................................................. 12 2.6.1 The transmission parts .......................................................... 12 2.6.2 The workshop ........................................................................ 12 2.6.3 The process planning ............................................................ 13
3 Problems to solve and research objectives ............... 15 3.1 Industrial problem to solve ............................................................ 15 3.2 Related process planning research .............................................. 16 3.3 Research objectives ...................................................................... 20 3.4 Delimitations .................................................................................. 20
4 Research approach ....................................................... 21 4.1 R&D of systematic process planning methods ............................. 21 4.2 Positioning of appended publications ........................................... 24
XII
5 Performed research and synthesis of results ............ 25 5.1 Schematic process planning ......................................................... 26 5.1.1 An approach for systematic process planning of gear transmission parts .................................................................. 27 5.2 Capture of process behaviour ....................................................... 28 5.2.1 Paper A: Process modelling using upstream analysis of manufacturing sequences...................................................... 30 5.2.2 Paper B: Analysis of process parameters during press quenching of bevel gear parts ............................................... 33 5.2.3 Multivariate data analysis ...................................................... 38 5.3 Tolerance synthesis and analysis ................................................. 39 5.3.1 Paper C: Tolerance chain design and analysis of in-process workpiece ............................................................................... 41 5.3.2 Paper D: Process chain based workpiece variation simulation for performance utilisation analysis ....................................... 44 5.4 Synthesis of results ....................................................................... 49 5.4.1 Information flows .................................................................... 49 5.4.2 Refining the PPP with information flows ................................ 52
6 Discussion and conclusions ........................................ 55 6.1 Coverage of the research results .................................................. 55 6.2 Follow-up of research objective 1 .................................................. 56 6.3 Follow-up of research objective 2 .................................................. 58 6.4 The relation between the PPP and PPAP ..................................... 58 6.5 Process capability index ................................................................ 59 6.6 Relevance of the chosen problems to solve ................................. 61 6.7 Scientific contribution ..................................................................... 62
XIII
6.8 General application of the research results .................................. 62 6.9 Conclusions ................................................................................... 63
7 Future work.................................................................... 65 7.1 Economic aspects ......................................................................... 65 7.2 Representation of the PPP and information flows ........................ 65 7.3 The model driven approach .......................................................... 66 7.4 A new approach for P-FMEA ........................................................ 66
8 References ..................................................................... 67
1
1 Introduction 1.1 Process planning for manufacturing Process planning for manufacturing covers a wide range of activities needed to specify the manufacturing process for a part. Unfortunately, the process planning terminology “is somewhat fuzzy” (Anderberg, 2012) and there is no commonly adopted or standardised definition of what is included or not, nor for the activities whatever they may be. Marri, et al. (1998) gives a definition that seems to be applicable in most cases where process planning is considered: “Process planning is defined as the activity of deciding which manufacturing processes and machines should be used to perform the various operations necessary to produce a component, and the sequence that the processes should follow.” A more comprehensive definition of the aim of process planning for manufacturing is that by ElMaraghy and Nassehi (2013): “Process planning, in the manufacturing context, is the determination of processes and resources needed for completing any of the manufacturing processes required for converting raw materials into a final product to satisfy the design requirements and intent and respect the geometric and technological constraints.” An important addition to these constraints is that the manufacturing must be cost-effective for a business to be successful. This is realised by a well-running manufacturing process in addition to efficient process planning to meet agreed deadlines (Scallan, 2003). The integration of manufacturing cost is essential and has been discussed by many authors (Gupta, et al., 2011; Xu, et al., 2009; Mayer & Nusswald, 2001). Process planning is also described to be the interface between product design and manufacturing (Scallan, 2003) and the work often includes coordination of product design intentions and constraints imposed by the workshop (Ham, 1988). The process planner therefore plays a desirable role not only to define a process plan but to contribute with manufacturing knowledge to a competitive product design (Bagge, 2009; Groche, et al., 2012; Inman, et al., 2013). To bring some clarity in different focuses for process planning, ElMaraghy and Nassehi (2013) have defined four process planning levels
2 | INTRODUCTION
according to Table 1. These levels are placed in order from a very low level of detail to a very detailed level. In addition, the focus and output from each level are identified. Table 1: Process planning levels (ElMaraghy & Nassehi, 2013). Process planning level
Main focus of planning at this level
Level of detail
Planning output at this level
Generic planning
Selecting technology and rapid process planning
Very low
Manufacturing technologies and processes, conceptual plans, and DFx analysis results
Macro planning
Multi-domain
Low
Routings, nonlinear plans, alternate resources
Detailed planning
Single domain, single process
Detailed
Detailed process plans (sequence, tools, resources, fixtures, etc.)
Micro planning
Optimal conditions and machine instructions
Very detailed
Process/Operation parameters, time, cost, etc., NC codes
This definition of process planning levels is established in the academic research community but is not so well known by process planners in industry. The probable reason is that in practice these four levels overlap each other and are difficult to distinguish. More easily recognised are the main activities related to process planning. These activities have been listed by Alting and Zhang (1989) in ten steps: 1. Interpretation of product design data (CAD data, material properties, batch size, tolerances, surface definition, heat treatment and hardness, special requirements). 2. Selection of machining processes (e.g. turning milling, drilling and grinding), usually based on a company specific strategy. 3. Selection of machine tools considering for example availability, process capability, machining range and production rate. 4. Determination of fixtures and datum surfaces. 5. Sequencing the operations. 6. Selection of inspection devices. 7. Determination of production tolerances.
INTRODUCTION | 3
8. Determination of the proper cutting tools and conditions (e.g. depth of cut, feed rate and cutting speed). 9. Calculation of the overall times (machining, non-machining and setup times). 10. Generation of process sheets, operation sheets and NC data. There is no established definition of what a process plan is, and what process planning work shall include, furthermore not all process planners in all companies may perform all of the activities. Process planning has been, and still is, mainly a knowledge intensive manual task performed by a skilled person because of its multiperspective problem nature including “problem-solving, constraintreasoning, goal-achieving, resource utilisation and conflict-resolution” (Ham, 1988). Although computer aided process planning (CAPP) has been on the production research agenda for decades it has not been commonly assimilated into industry (Xu, et al., 2011; Anderberg, 2012). In the end, the goal for the process planner is to design a manufacturing process that is capable of producing a part that fulfils all design requirements, without incurring high economic costs. Xu, et al., (2009) emphasise the major impact that process planning has on factory activities and the resulting manufacturing costs. Process planning is, and will be, an important issue for all manufacturing companies, no matter whether manual, computer-aided or, most probably, a combination.
4 | INTRODUCTION
1.2 Structure of the thesis 1. Introduction -A common description of process planning for manufacturing
2. Frame of reference -Process planning as considered in this thesis
3. Problems to solve and research objectives 3.1 Industrial problem to solve
3.2 Related process planning research 3.3 Research objectives
4. Research approach 5. Performed research and synthesis of results Objective 2
Objective 1 Framework of process planning methods
Licentiate thesis Paper A Paper B
Paper C and D
PPP
Refined PPP Part design
Part design F G
F
Assignment directive
Schematic process plan
Schematic process planning
D
C
NO
Initial process planning
In-process tolerances
Process plan concept
Tolerance analysis
Tolerancing In-process dimensions
Predicted outcome of process
YES
D
Good enough?
Establish process plan
Process plan!
I Defining process control strategy
Process control strategy
K K
A
H
Examples of supporting processes to create:
G
Work instructions
B Simulations
E
Control documents
Product characteristics
Tool layout NC-programs Measuring programs
Process behaviour In-process tolerances
Design of Experiments
...
Process settings
Multivariate data analysis
I H
H J Collaboration parties: Workshop Manufacturing technology
6. Discussion and conclusions
7. Future work Figure 1: Structure of the thesis.
5
2 Frame of reference The overall scope of this thesis comprises all levels of process planning described in Table 1 but does not deal in detail with all potential topics and activities. The following will be an introduction to the frame of reference in which the research has been performed and to the origin of and ideas for resolving the subsequently identified problems. 2.1 PPP - the process planning process To clarify the general conception of process planning it is appropriate to make some more statements and explain the desired goal for the process planner. Process planning is work that includes all needed activities to define a process plan. A process plan is a specification that contains information aimed to facilitate production of a certain part in a definite manufacturing system. In practice, two tacit boundary conditions narrow this definition; the process plan must be defined in a way that ensures that the produced part fulfils all requirements stated in the design specification as well as minimising the production cost. As process planning in general is a very broad topic it is here delimited to fit the purpose of this thesis. My approach to giving an overview of the process planning domain is shown in Figure 2 as a flow chart of the “process planning process” (PPP). The PPP flow chart includes activities derived from my own experiences and ideas; many of them, especially sub-activities, are also found among the ten activities listed by Alting and Zhang (1989).
6 | FRAME OF REFERENCE
Part design
Part design
Schematic process planning
Schematic process plan
Assignment directive NO
Initial process planning
Process plan concept
Tolerancing
Defining process control strategy
In-process work piece tolerances
Tolerance analysis
Result
Good enough?
YES
Establish process plan
Process plan!
Process control strategy
Examples of supporting processes to create:
Simulations
Process behaviour
Work instructions Control documents Tool layout NC-programs
Design of Experiments
Process behaviour
Multivariate data analysis
Process behaviour
Known process behaviour
Measuring programs ...
Figure 2: The PPP flow chart – an approach to describe the process planning process (PPP).
According to Figure 2, the main flow starts with the activity “initial process planning” on receiving an assignment directive. The result is a process plan concept where the main directions of the following work have been set. This is used together with knowledge about the behaviour of relevant manufacturing processes as input to the three following process planning activities. These three activities are interdependent and are performed in an iterative way starting with schematic process planning. Schematic process planning includes, for example, interpretation of design requirements, definition of production and operation sequence, machine tools, cutting and work holding tools.
FRAME OF REFERENCE | 7
Tolerancing is to define in-process workpiece (IPW) tolerances in conjunction with the schematic process plan, process behaviour and design specification of the part. The process control strategy is based on both the schematic process plan and the required tolerances, but also on the behaviour of all included processes. The defined process plan is then analysed regarding tolerances and the expected outcome of the process. The results are tested against acceptance criteria to decide whether the defined manufacturing process can be approved or not. If the manufacturing process has all the necessary qualities to succeed, the process plan can be established and all needed documents, programs, working instructions, etc. created. 2.2 Process capability – a boundary condition for process planning Because a manufacturing process that produces virtually no products out of specification tends to be very expensive, there is often an acceptance criterion allowing a small amount of deviation from specification. This criterion is commonly defined as the lowest acceptable level of a process capability index 1 (PCI), the highest number of defect parts per million (PPM) or sometimes as maximum defects per million opportunities (DPMO) (Wu, et al., 2009). The aim of using this kind of criteria is to prohibit poor production process performance in relation to the requirement, preventing adherent high quality deficiency costs. PCIs is often used to set a limit for the lowest acceptable process capability, but has not been used for limiting the highest acceptable process capability. The aim of a higher limit should be to prevent excessive manufacturing cost due to overqualified manufacturing resources; this is in contrast to the commonly used lower limit which aims to save the customer from getting a product out of specification. Even though a useful higher limit of capability index is not defined in the literature, the assumption here is that the process planner must try to obtain a balance between required product performance and the economic efforts put into the manufacturing process.
1
Capability index (typically Cp, Cpk, Cm, or Cmk) represents the relation between the requirement (tolerance range) and the statistically defined process behaviour (e.g. 6*standard deviation). (Montgomery, 2009)
8 | FRAME OF REFERENCE
2.3 Conceptions of process characterisation Until now, three conceptions of a similar kind have been used to characterise a process; process performance, process behaviour and process capability. Some clarifications are in order to avoid confusions. “Process performance” refers to the accuracy of the process when creating a final part dimension or an IPW dimension. “Process behaviour” is a representation of the manner, or action, of the manufacturing process. “Process capability” is dedicated to the established conception of PCI where a tolerance range is related to a statistical representation of the process behaviour. 2.4 The beam balance My conceptual approach to show the effect of the relation between product performance and production performance is illustrated by the beam balance in Figure 3 The leftmost scale pan contains a load that represents the product performance, stated as requirements in the design specification. The rightmost pan contains the load of the manufacturing process performance as a consequence of the process plan.
High cost!
Competitive cost?
a)
b)
High cost!
c)
Figure 3: A balance beam balancing product performance and manufacturing performance.
FRAME OF REFERENCE | 9
If the decided manufacturing process cannot balance the product design requirements, as shown in Figure 3 a, there will be a situation with more out-of-specification parts. If the effect of this poor quality is managed within the production system, for example by putting extra efforts in measuring, sorting procedures, re-work and scrapping, the production cost is assumed to be high (Fischer, 2011). If the poor quality is not given enough attention and the parts reach a down-stream customer, like the assembly line, the costs tend to be very high and even worse, if the final customer recognises parts out-of-specification, there is a risk of extremely high expenses. The inverse situation (Figure 3 b) occurs when the manufacturing process is specified to contain resources where the capabilities exceed the requirements. The production cost is then assumed to be higher than necessary because of unutilised performance of the manufacturing resources. In addition, there is a risk of increased processing lead times when using manufacturing equipment with unnecessarily high quality capabilities (Mayer & Nusswald, 2001). This is a contributory cause for high overall production cost. 2.5 The beam balance versus process capability To elaborate the rather intuitive approach with the balance beam, I have created a simplified causal graph model in Vensim2, shown in Figure 4. This model includes the necessary underlying factors and the relations to understand the balance beam and show why process capability can be regarded as the other side of the same coin.
2
Vensim is software for “system dynamics” modelling. System dynamics is a methodology for modelling complex, dynamic casual systems (Sterman, 2001). Vensim is also used later on to simulate manufacturing processes (Bagge, et al., 2014).
Sales price
10 | FRAME OF REFERENCE
Product performance 1/x
Product performance
Manufacturing process variation
=
1/x
Manufacturing process performance
Capability
Cost of poor quality
CoPQ
Tolerance width
Process capability
+
External
-
Lookup CoPQ
Earnings
-
Internal Manuf. resource cost
Balance
Sales price Lookup price
Manufacturing resource cost Manuf. process performance
+
+
Manufacturing cost
Lookup manuf cost
Figure 4: Relations between product performance and manufacturing process performance.
According to the right part of Figure 4, earnings when selling a product are defined as the difference between the manufacturing cost and the sales price. Sales price is highly dependent on the product performance aimed to satisfy the customer 3. Manufacturing cost is a result of both manufacturing resource cost and, if any, an internal cost of poor quality (CoPQ). Manufacturing resource cost depends on the manufacturing process performance. This relation is most frequently illustrated as a chart showing possible tolerance width versus the manufacturing cost (left part of Figure 5) (Manoharan, et al., 2012; Hamou, et al., 2006; Wu, et al., 1998). An increasing tolerance width results in lower manufacturing cost thanks to a decreased need for manufacturing process performance. This can correspondingly be expressed as shown in Figure 5 b, where increased manufacturing process performance drives higher manufacturing cost.
3
The used curve in Figure 4 is just an attempt to illustrate a conceivable relation between sales price and product performance.
Manufacturing cost
Manufacturing cost
FRAME OF REFERENCE | 11
Tolerance width
Manufacturing process performance
Figure 5: Manufacturing cost versus tolerance width (left) and versus process performance (right).
The cost of poor quality in Figure 4 (Anderberg, 2012) is a result of a low-capability manufacturing process that (too often) produces parts out of tolerance. If the CoPQ appears internally, before delivery to the final customer, it is here regarded as an undesirable contributor to the manufacturing cost. If the customer is subjected to poor quality, the CoPQ (external) will have a negative impact on earnings, no matter whether it is put on the manufacturing account or elsewhere. Capability originates from the relation between the tolerance width and the manufacturing process variation and drives CoPQ if it is too low. Tolerance width and manufacturing process variation are in turn the inverses of product performance and manufacturing process performance respectively. The quotient is a measure of the balance between these two performances. This implies that the beam balance approach can be supported by the use of PCI as an indicator for what is a well-balanced manufacturing process design. In the short term, imbalances between the product design and the production facilities can be accepted. If the manufacturing process capability is low, there may be a temporary solution to filter off bad quality parts by additional measuring and checking (Anderberg, 2012). If there are high performing resources available in the workshop, they may be used in favour of investments in new, less sophisticated equipment. As the production efficiency, final cost and product quality are affected by the process plan (Chang, 2005), the desirable balance between product performance and manufacturing process performance (Figure 5 c) is definitely considered to be dependent on the work of the process planner.
12 | FRAME OF REFERENCE
2.6 Transmission part manufacturing As already explained, the work of a process planner cannot be unambiguously defined because it will contain different activities, strategies and methods depending on situation. Typically, the type of part, production rate, annual volumes, company strategies, etc. gives different possibilities and constraints for process planning. The following will give an insight into the area of industrial transmission part manufacturing from which process planning experiences and the research presented in this thesis are derived. 2.6.1 The transmission parts The studied company both develops and manufactures transmission parts in-house. Precision manufacturing is performed in workshop facilities aimed at series production of gear wheels, pinions, gear box shafts, etc. for heavy trucks, busses and coaches. Annual production volumes for each category can be counted in terms of ten thousand to a few hundred thousand parts. Each part is usually in production for at least a couple of years. Figure 6 shows cylindrical gears for a gearbox and bevel gears for rear axles.
c) a)
d) b) Figure 6: Examples of transmission parts: Bevel gear pinion (a), bevel gear ring gear (b), counter shaft (c) and gear wheel (d).
2.6.2 The workshop The manufacturing process for these kinds of parts can be divided into three sub-processes; 1) soft machining, 2) heat treatment and 3) hard machining, as shown in Figure 7.
FRAME OF REFERENCE | 13
Raw material /blanks
Soft machining
Turning
Drilling
Gear cutting
Heat treatment
Carburiz- Quenching ing
Tempering
Hard machining
Turning/ grinding
Honing
Gear grinding
Inspection and measuring
Figure 7: A typical manufacturing process chain for gears.
These three sub-processes contains various multi-step machining or heat treatment operations organised on different kinds of manufacturing lines. A common line for soft machining contains one or two multispindle lathes, machine tools for gear and spline cutting and some equipment for chamfering/deburring. Heat treatment is normally case hardening followed by tempering. Hard machining lines usually include machine tools for hard turning, grinding, honing and gear grinding. Depending on transmission part category, a complete manufacturing process includes 6 to 12 set-ups, measuring excluded, for the part to undergo before it is finished. Each line is designed and equipped to produce one kind of transmission part such as gear box gears, gear box main shafts, gear box counter shafts, bevel gear pinions, etc. Depending on part design, the fixtures, cutting tools and NC programs are changed to facilitate manufacturing of variants within the desired operational range of the line. 2.6.3 The process planning As a consequence of the product design, most transmission parts are able to be grouped or categorised, and the process plans are more or less systemized in accordance with these part categories. For example, bevel gear sets for the rear axle (one pinion and one ring gear) may have different gear ratios depending on customer needs, but the gear housing axle casing is the same. This is realized by using standardised interfaces between gears and housing and then defining a range of gear sets that will
14 | FRAME OF REFERENCE
fit. These standardised interfaces are utilised also in the manufacturing process. All the pinions are manufactured in the same way, on standardised machining lines with the same kind of machine tools. Most of the cutting tools and fixtures are also the same, but with some specific NC programming and gear cutting tools suited to the individual, gear ratio dependent dimensions. This structure is emphasised by the company strategy, advocating both systematic part design and manufacturing process design, which in turn characterises the process planning work. The process planning methods can be regarded as mostly retrieval because of this strategy. This implies that the process planning has more of a continuous improvement and long-term perspective than handling completely new product designs in a workshop with frequent changeovers. Much effort can be put into getting proper designs for both the manufacturing process and the part. This industrial production environment is the cause and trigger of many thoughts and ideas of methods for systematic planning of precision manufacturing processes.
15
3 Problems to solve and research objectives Two main process planning problems are discussed in this thesis. The first has its background in my own industrial experiences and relates to the need for different kinds of process planning methods. The other is identified as a result of studies of related academic research about process planning. The research objectives are a consequence of these two problem definitions as illustrated in Figure 8.
3. Problems to solve and research objectives 3.1 Industrial problem to solve 3.2 Related process planning research 3.3 Research objectives
Figure 8: The research objectives are derived from both an industrial and an academic problem identification.
3.1 Industrial problem to solve Process planning is a challenge because of the diversity of activities required to design and evaluate a manufacturing process. Human knowledge has to be involved, yet it is not possible to efficiently transform this into CAPP systems because of the multi-perspective problem nature. Because many people are usually involved in product and manufacturing process development much process planning information has to be both easily accessible and comprehensible for others than the process planner. Tacit process planning knowledge and reasons for decisions have to be explicitly described. The first issue when developing methods for process planning is to facilitate systematic representation of qualitative information. The other issue is to integrate methods for quantitative process planning information. Product design specifications are with only few exceptions based on quantitative measures. Whatever the kind of acceptance criterion for good quality is and what is regarded as an acceptable production cost, there is a need to estimate the outcome of a proposed manufacturing
16 | PROBLEMS TO SOLVE AND RESEARCH OBJECTIVES
process. The most decisive question before finalizing a process plan is whether the outcome of the manufacturing process will be as expected. Will the parts comply with the specification? How often will an out-oftolerance part appear? The process planner must quantify and ascertain the “truth” about the production system capabilities to be able to answer these questions and define a process plan that ensures well-running production. Even if the produced parts fulfil the specifications, the manufacturing process is not automatically well-utilised in terms of production system capabilities. The mission for the process planner is to design the process in a way that gives a good yield in terms of required quality but also low cost thanks to a smooth running production with well utilised workshop resources that are suited to the purpose. In multi-step manufacturing processes, which are used for gear transmission parts, there are many concurrent parameters that must be considered and the process planner has many choices to make while developing the process plan. The process plan must be more or less detailed to avoid deviations due to variations in machine tool set-up, cutting tool change, process control strategy, etc. There will often be a broad range of possible solutions for bringing different process steps together and define for example a control strategy for a complete manufacturing process. There is a need to analyse the effect of potential process plan solutions, parameter settings, IPW tolerances, control strategies, etc. and there has to be one or more methods available where these aspects can be defined and tested. Predicting manufacturing process outcome and making comparisons to the design part specification have to be integrated parts of a framework of methods to support process planning. There is a need for a framework of process planning methods integrating both descriptive/qualitative and analytic/quantitative parameters and able to reveal whether the proposed process plan is good enough. 3.2 Related process planning research An introduction to the nature of process planning and how it is defined in literature has been given in chapter 1. The following gives further details of previous research efforts on topics related to the identified problem.
PROBLEMS TO SOLVE AND RESEARCH OBJECTIVES | 17
The greater part of modern process planning research has been devoted to CAPP. CAPP has challenged the manufacturing research community for the last four decades and a huge volume of literature has been published (Xu, et al., 2011). Though there are extensive efforts put into CAPP research, and many different CAPP systems have been developed, there is still a long way to go before process planners can commonly use the technique. This was the overall conclusion drawn 2025 years ago about the state of CAPP research and implementation (Alting & Zhang, 1989; ElMaraghy, 1993) and the situation is hardly changed since (Xu, et al., 2011; Anderberg, 2012). The problem with CAPP is characterised by many interdependent parameters and variables (Denkena, et al., 2007), not only technical but economical, organisational and environmental. However, this problem is not peculiar to CAPP but characterises the complex and dynamic nature of process planning in general (Ham, 1988), whether it is performed by computers or humans. As the process planner spends a great deal of time on creating and administrating operation sheets, drawings, text documents, etc. (Lundgren, et al., 2008; Halevi & Weill, 1995) there is an incentive to rationalise and automatise such work. This is one of the topics that CAPP research has been focusing on and what can be associated with research areas such as product lifecycle management (PLM), computer-aided design (CAD) and computer-aided manufacturing (CAM). As indicated in the PPP (Figure 2) these kinds of activities are done when a process plan is established and most of the process planning decision-making is already done; for example, there are available definitions of operation sequence, suitable machine-tools, tolerances and control strategies. This thesis has its focus on the activities leading to establishing the process plan according to the PPP, i.e. process planning decision-making, roughly described as manufacturing process design. Here, the CAPP research has been devoted to different approaches to reflect the heuristic character of human-based process planning and develop computer systems showing some traits of human intelligence. Artificial intelligence (AI) has the goal of modelling human intelligence on computers, and has been one of the main approaches for automated process planning. Rzevski (1995) has listed five paradigms of AI in engineering: knowledge-based systems, neural networks, genetic algorithms, fuzzy logic and intelligent agents. Most of the AI-related CAPP research can be
18 | PROBLEMS TO SOLVE AND RESEARCH OBJECTIVES
classified according to these paradigms with examples given in Table 2. Some of them are hybrids covering more than one paradigm. Table 2: Examples of research in different paradigms of artificial intelligence (AI). Paradigm of AI
Examples of research
Knowledgebased systems
(Kusiak, 1991; Xu, et al., 2009; Zhang, et al., 2012), including definition, storage and retrieval of relevant process planning information, often incorporated in a PLM system (Denkena, et al., 2007)
Neural networks
(Teti, et al., 1999) (Hua & Xiaoliang, 2010) (Amaitik & Kiliç, 2007)
Genetic algorithms
(Qiao, et al., 2000) - operation sequencing (Cai, et al., 2009) integrates process planning and scheduling (Li, et al., 2002) (Azab & ElMaraghy, 2007) Reconfigurable process planning (Nejad, et al., 2012) Model based tolerance analysis
Fuzzy logic
(Amaitik & Kiliç, 2007) (Chen, et al., 1995) (Li & Gao, 2010) Integrates process planning and scheduling
Intelligent agents
(Li, et al., 2011) (Bose, 1999) Operation sequencing
Much of the AI–based CAPP research has an ambition to develop systems that automatically optimise the process plan, whether it is about operation sequencing, scheduling, tool selection, tolerancing or some other task. Many of these systems are shown to be suitable and well working for a highly delimited environment with well-defined constraints, but are not up to the wide-ranging characteristics of a complete process planning scope. Existing CAPP solutions do not have the possibility to include for example business and strategic considerations, which often have an impact on the choice of technology and machining resources (Denkena, et al., 2007). Finding research with the emphasis on comprehensive and heuristic process planning methods and understanding is difficult. Although, Timm et al. (2004 a, 2004 b) are good examples of where a heuristic and graphical approach is used for developing a mathematical model to support and link design dimensioning and process planning. Some general design rules can then be derived and explained, based on the model and the results.
PROBLEMS TO SOLVE AND RESEARCH OBJECTIVES | 19
Even more focus on the working processes, without the ambition of AI, self-learning abilities or automatic optimisation, is a qualitative and knowledge based approach to manufacturing process planning presented as a part of “Produktionslotsen” (DMMS, 2009) (in English; the production pilot) (Chen, et al., 2011). Produktionslotsen contains systemised, model-based guides to support not only process planning but other production-related activities such as factory planning, production investments and continuous improvements. A platform for model driven process planning for machining is presented by Hedlind (2013). With this platform, not only the human perspectives of visualisation and interaction of models can be satisfied but also the technological aspects of standardised information representation and exchange in a product realisation environment. Probably, it is this kind of comprehensible approach, with the potential to gather process planning intelligence, knowledge and information to support efficient decision-making, that has to be developed to bring CAPP one step towards industrial implementation. In the CAPP review (25 years ago) Ham (1988) pointed out the importance of detailed understanding of the task to be automated: “In that sense, before pursuing any computer-based effort toward process planning, it is necessary for us to move one step back and ask ourselves: do we really understand the process planning task well enough for computer automation? In other words, the fundamental questions such as what is process planning, how is it being done presently in actual practice, are there more logical ways to doing it must be answered before any programming efforts should take place. Given even the current state of computer technology, including artificial intelligence, it is still not possible to develop computer programs to intelligently carry out tasks unless we, as intelligent human beings, fully understand those tasks.” In spite of the fact that computer technology has experienced huge developments since 1988, lack of process planning understanding and systematisation still seems to be a reason that restricts the relevance of CAPP research and the successful industrial implementation.
20 | PROBLEMS TO SOLVE AND RESEARCH OBJECTIVES
3.3 Research objectives Two research objectives are defined as a consequence of the identified problems in sections 3.1 and 3.2. Research objective 1 The first objective has an all-embracing aim of finding methods that cover essential activities for process plan synthesis and analysis, including the possibility to predict the outcome of a proposed process plan. These essential activities are all found in the PPP flow chart in section 2.1. Much attention is paid to understanding and analysis. Research objective 2 The second objective is a result of the request for systematisation and a deeper understanding of process planning, addressed by the academic production research community. The aim here is to develop the PPP flow chart on the basis of experiences and findings from the methodology studies addressed in objective 1. 3.4 Delimitations The last activity in the PPP flow chart “Establish process plan” is not covered by the performed research; i.e. creating documentation required for production, as operation sheets, control plans and NC programs. Focus is on the underlying process planning efforts needed for its creation. Even though the manufacturing cost is an important factor it is not included in this thesis other than in general terms as described in section 2. “Process performance” is just a matter of geometric dimensions, it does not include any aspect concerning productivity, production rate or volume capacity, for example. Time related aspects such as machining cycle time or lead time through a complete manufacturing process are not considered. This thesis does not go into automatic optimisation procedures to find for example appropriate quality or balanced tolerances. Production scheduling and planning is not treated, nor process planning for job-shops.
21
4 Research approach The origin of the work presented in this thesis is years of production engineering practice for the manufacturing of transmission parts such as gears, shafts and synchromesh. Experiences and ideas are mainly derived from my own process planning work, performed in an industrial environment where the tasks have been both managing and improving mature processes and products in addition to development of new ones. The research has been characterised by adopting a methodology based approach to meet the objectives stated in section 3.3. The approach is comprehensive and aims to gather essential methods that should cover the process planning domain as shown in Figure 2. The level of details is not restricted and can include all four levels defined by ElMaraghy and Nassehi (2013) in Table 1. Strategies for defining tolerances and evaluation of manufacturing process chains are examined with most of the focus on the detailed and micro-planning levels. The purpose of examining these strategies is to facilitate a well-substantiated answer to the final revealing question: -Is the process plan good enough? The work is also related to the request for more process planning knowledge addressed by Ham (1988) in chapter 3.2. The approach to meet this need is to refine the PPP by using results derived from the methodology studies. The research is not aimed at completing any existing CAPP approach but to independently contribute to better process planning understanding, without limitations in available computer-based process planning aids. 4.1 R&D of systematic process planning methods The presented research extends the qualitative approach for systematic process planning (Bagge, 2009) to include quantitative and analytical methods for design and evaluation of a process plan. A conceptual illustration of the research and development (R&D) process for this work is shown in Figure 9.
22 | RESEARCH APPROACH
As is
To be
Schematic process planning
Schematic process planning
Complete process planning
Descriptive, qualitative
Descriptive, qualitative
Descriptive, qualitative Analytic, quantitative Revealing
New research Capture of process behaviour
Capture of process behaviour
Perceptual, qualitative
Quantitative (Perceptual, qualitative)
Figure 9: Research and development of systematic process planning methods.
The licentiate thesis “An approach for systematic process planning of gear transmission parts” (Bagge, 2009) covers descriptive and qualitative methods for process planning, e.g. how to deal with product design requirements, part and manufacturing datums, manufacturing operations and manufacturing sequence, all with an explanatory approach. The left part of Figure 9 represents this schematic process planning and the tacit, but crucial, knowledge about process behaviour. However, making analyses and predictions of the process plan requires quantification both of the requirements, typically manufacturing tolerances, and the behaviour of the included manufacturing processes. The purpose of the new research is to reach the complete process planning state , “To be”, on the right of the arrow in Figure 9, where not only descriptive and qualitative methods are defined but also analytic and quantitative methods. This part of the work corresponds to research objective 1. Research objective 2 is realised by using results derived from the work for objective 1 and has not been published separately. Research objectives 1 & 2 and their relation to the performed research are showed in Figure 10.
RESEARCH APPROACH | 23
3. Problems to solve and research objectives 3.1 Industrial problem to solve
3.2 Related process planning research
3.3 Research objectives 1. Find suitable methods for process planningi 2. Increase knowledge about process planning
4. Research approach
5. Performed research and synthesis of results Objective 2 Objective 1 Framework of process planning methods
Licentiate thesis Paper A
Paper B Paper C and D
PPP
Refined PPP Part design
Part design F G
F
Assignment directive
Schematic process plan
Schematic process planning
D
C
NO
Initial process planning
In-process tolerances
Process plan concept
Tolerance analysis
Tolerancing In-process dimensions
Predicted outcome of process
YES
D
Good enough?
Establish process plan
Process plan!
I Defining process control strategy
Process control strategy
K K
A
H
Examples of supporting processes to create:
G
Work instructions
B Simulations
E
Control documents
Product characteristics
Tool layout NC-programs Measuring programs
Process behaviour In-process tolerances
Design of Experiments
...
Process settings
Multivariate data analysis
I H
H J Collaboration parties: Workshop Manufacturing technology
Figure 10: Research objectives 1 & 2 related to the performed research.
24 | RESEARCH APPROACH
4.2 Positioning of appended publications The appended publications in this thesis and their coverage are mapped onto the PPP flow chart as shown in Figure 11. Experiences from the FFI ReVer project have not yet been published but are partially described in this thesis as an example of multivariate data analysis.
Part design
Part design
Licentiate thesis (Bagge 2009)
Schematic process planning
Schematic process plan
Assignment directive NO
Initial process planning
Process plan concept
Tolerancing
In-process work piece tolerances
Tolerance analysis
Result
Good enough?
YES
Establish process plan
Process plan!
Paper C (Bagge et al. 2013) Defining process control strategy
Paper A (Werke et al. 2014)
Process control strategy
Paper D (Bagge et al. 2014)
Examples of supporting processes to create:
Simulations
Process behaviour
Work instructions Control documents Tool layout NC-programs
Paper B (Bagge Lindberg 2012)
Design of Experiments
Process behaviour
FFI ReVer project
Multivariate data analysis
Process behaviour
Known process behaviour
Measuring programs ...
Figure 11: Coverage of appended publications upon the PPP flow chart.
25
5 Performed research and synthesis of results The five publications shown in Figure 11 are turned to account in this thesis, and all discuss supporting methods for process planning. These methods have been gathered into a framework for process plan development shown in Figure 12 as a result of the research undertaken. This framework is introduced here to provide an overview and better understanding while reading the following summaries of the publications.
Illustration of part geometry Comparison
Product dimensions & tolerances
Representation of product dimensions
Mapping matrix
PS
PS
PS
view of IPW variation
Representation of IPW dimensions
IPW dimensions & tolerances
OE
OE
OE
view of IPW variation
Representation of IPW dimensions
FD Process chain dim. & tol.
Mapping matrix
IPW dimensions
3 1
2
Finished part
C
b
A
B
a
Turning A’
b’ End-machining and B’ centre drilling
b’
a’
a’
Face driver
Retractable jaws
Parting line and residual flash
Figure 12: Framework of methods for process plan development.
The framework contains three vital process planning subjects indicated in Figure 12: 1. Schematic process planning 2. Capture of process behaviour 3. Tolerance synthesis and analysis
26 | PERFORMED RESEARCH AND SYNTHESIS OF RESULTS
Each subject has one or more supporting methods to realise the subject intentions, all outlined in this thesis. Subject 1: Schematic process planning The methodology for systematic process planning covers interpretation and transfer of design requirements into the schematic process plan. The schematic process plan facilitates explanatory representation of machining or heat treatment operation elements, process sequence and important decisions such as the choice of datum, clamping surfaces and tools. The representation of measuring operations and measuring equipment is done by the same principles. It is desirable to highlight key features derived from both design specification and manufacturing process development. Subject 2: Capture of process behaviour Subject 2 contains the supporting methods; simulation, design of experiments (DOE) and multivariate analysis. The aim is to capture knowledge and information about the behaviour and abilities of potential manufacturing processes to be included in the process plan design. Subject 3: Tolerance synthesis and analysis The DDC2 technique is used to support the development of a process plan regarding synthesis and analysis of tolerances. Different process plan solutions, tolerancing strategies and control strategies can be tested and the outcome of a contemplated or existing manufacturing process predicted. In the following sections, summaries of the appended publications are presented after a short introduction to each subject. 5.1 Schematic process planning The schematic process planning has been characterised as a descriptive and qualitative part of the process planning domain. This has mainly been dealt with in the licentiate thesis “An approach for systematic process planning of gear transmission parts” (Bagge, 2009).
PERFORMED RESEARCH AND SYNTHESIS OF RESULTS | 27
5.1.1 An approach for systematic process planning of gear transmission parts The licentiate thesis (Bagge, 2009) introduces the field of process planning of transmission parts, based on experiential research. What role does the process planner play in different situations, what is the area of responsibility, what motivates the use of systematic and transparent process planning methods? Figure 13 characterises the interaction between the process planner and collaboration parties while developing a process plan.
Process Plan: Methods
Part design
Operation sequence Machining data Tolerances
Workshop facilities
Cutting tools Clamping tools
Measuring strategy
Manufacturing technology
Figure 13: The role of the process planner.
There is always an inflow of information from part design to process planning, and often a possibility to feed back to part design. This depends on the situation and where the designed part is in the product life cycle. In product development stages it is common to have collaboration between part design and process planning and here are considerable possibilities to make design adjustments. When the product is mature, possibly becoming a spare part, and a sub-supplier is assigned to start production, the threshold for changing the design is conversely high. In addition, collaboration with external providers of manufacturing technologies very much depends on the situation and on company strategies. Often process planning is based on experiences and knowledge of the process planner, performed by intuition, and without any explicit
28 | PERFORMED RESEARCH AND SYNTHESIS OF RESULTS
procedures for working. The only documentation is the finally defined process plan and the traceability of made decisions is poor. The licentiate thesis proposes a method that facilitates visualisation of the intentions of the process planner, the prerequisites and the circumstances for made decisions. This makes the determining factors for the resulting process plan transparent and comprehensible. One important aspect of process planning is tolerancing, which must be considered to complete the definition of how a part shall be manufactured. Detailed tolerancing is not included in the licentiate thesis but the principles of IPW tolerance chains (tolerance stack-ups) are pointed out. Identification of relations between machining datums and created features is described by using a datum hierarchy (rooted-tree) diagram. Bagge ends up with the following conclusions about the proposed method for systematic process planning (Bagge, 2009): It describes a systematic approach to interpret new as well as mature products. It prescribes systematically how essential design characteristics should be transferred to the process plan. It points out how attention should be paid to datums It guides the process planner through the evaluation of operations It proposes a way to document the process planning work, aimed at both immediate and future needs. Further performed research and publications are along the same line as the proposals expressed in the licentiate thesis; tolerancing, quantifying performance of needed manufacturing processes and capability testing of the developed process plan. 5.2 Capture of process behaviour One precondition for creating a relevant process plan is to have knowledge about the abilities and behaviour of the processes to be used. This is essential when IPW tolerances and control strategy shall be defined. Depending on the situation and the exactitude of the process planning work, different degrees of understanding of process behaviours are required. It is possible, but not always advisable, to compensate for deficient information about process abilities by defining (too) wide IPW
PERFORMED RESEARCH AND SYNTHESIS OF RESULTS | 29
tolerances. This will result in low utilisation of the process performance. If the process behaviour does not correspond to what is assumed, and the IPW tolerances are too tight, there is an obvious risk of difficulties to control the process and, as a result, getting parts out of specification. Knowledge of manufacturing process behaviour can be captured in many ways, in different situations and for different purposes. This thesis identifies and deals with three different methods according to Figure 14 for investigating the behaviour of complex processes. One or more of these methods can be used to shed light on process behaviour and bring valuable input data to the development of the process plan.
Simulations
Process behaviour
Design of Experiments
Process behaviour
Multivariate data analysis
Process behaviour
Known process behaviour
Figure 14: Methods to capture process behaviour.
When developing a new process plan for a new part, with new and unrealised manufacturing processes, there will at the start be little or no available knowledge about process capabilities. This implies the use of methods for simulation to get one step closer to what is required. Section 5.2.1 refers to Werke, et al. (2014) who proposes a methodology for process modelling of manufacturing sequences. If there is a need for new, or more detailed, information about especially complex processes, DOE has been proven to support the process planner (Bagge & Lindberg, 2012). This is further referred to in section 5.2.2. For established processes in workshops specialised for production of high volumes of the same kind of parts, there is often a lot of production data available. Work package A of the Realistic Verification (ReVer) project has the intention to use available historical process and material
30 | PERFORMED RESEARCH AND SYNTHESIS OF RESULTS
data to predict geometrical distortion of press quenched gears. A tool based on data analysis should then support the process planner. This kind of multivariate data analysis approach is discussed in section 5.2.3. 5.2.1 Paper A: Process modelling using upstream analysis of manufacturing sequences Simulations enable the examination of processes not only in the initial development stages, but also when they are well established. One reason for making simulations of established processes is to facilitate efficient process start-up and control. For example, when material properties have a significant impact on a machining or heat treatment operation, proper process settings can be foretold by simulations. Werke, et al. (2014) takes a step closer to efficient modelling and simulation of manufacturing processes by proposing an algorithm for upstream process modelling. The methodology defines a set of rules for system modelling of manufacturing sequences to be used in process planning. It facilitates the selection of critical process parameter and an evaluation of how variations in these parameters influence the final component properties. The virtual manufacturing chain acts as an important source for extracting and quantifying the process behaviour that is needed when developing and controlling a process. In contrast to the more intuitive downstream approach, the upstream approach is governed more by the purpose of the simulations; not to model a complete manufacturing chain and then make fragmentary simulations. This is illustrated in Figure 15 where a conceptual framework of models for making simulations to evaluate one or more properties of the final part is defined. In this example, it is obvious that only some of the process steps in the physical chain are represented in the virtual chain. The reason is that only process steps affecting the finally examined properties are modelled. The principles of the upstream process modelling are along the same line as the method for upstream process planning (Bagge, 2009) illustrated in Figure 16.
PERFORMED RESEARCH AND SYNTHESIS OF RESULTS | 31
Physical chain: Process param.
Process param.
Process param.
Process param.
Process param.
Process param.
Process param.
Process step 1 (First)
Process step 2
Process step 3
Process step 4
Process step 5
Process step 6
Process step 7 (Final)
Process param.
Process param.
Sim. 6,1
Sim. 7,1
Final workpiece
Framework of simulations: Process param.
Model data Current process
Model data Current process
Model data Current process
Sim. 4,1
Process param.
Process param.
Process param.
Sim. 5,1
Sim. 2,1
Sim. 4,2
Process param.
Process param.
Sim. 2,2
Sim. 5,2
Accumulated results
Figure 15: Transformation of a physical manufacturing sequence into a framework of models and simulations.
Manufacturing Process Chain
Final operation
First operation
Process Planning
Figure 16: Process planning methodology compared to manufacturing sequence.
32 | PERFORMED RESEARCH AND SYNTHESIS OF RESULTS
The process planner starts with the specification of the finished part and defines the final manufacturing operation first. The work then proceeds by defining required upstream process steps to fulfil the final part specification. Two cases in the paper exemplify the upstream process modelling methodology; one for a bevel gear pinion and one for a forged steering arm. The purpose of the virtual chain for the pinion shown in Figure 17 is to simulate and reduce the accumulated displacements after case hardening and in the end eliminate a straightening operation.
Figure 17: Framework of simulation activities for the pinion case study.
In conclusion, the paper proposes an algorithm to establish simplified metamodels of manufacturing sequences using breadth first search. The algorithm is based on stepwise upstream selection of process steps and facilitates extraction of a virtual simulation sequence from a physical manufacturing sequence. The virtual sequences can be used for optimisation of process parameters and evaluation of the effects of replacing, removing or adding process steps to a manufacturing sequence.
PERFORMED RESEARCH AND SYNTHESIS OF RESULTS | 33
Additional comments on the results in paper A Even if each simulation case has its own focus and well-defined demand for output data, there is a potential to derive more useful information from simulations by using the same, or slightly modified, models but ask for other characteristics. For example, the pinion case has its focus on geometric displacements, but the simulation models used also contain information about steel properties like carbon content, and residual stress distributions (Werke, 2009). This information may be of interest for the product designer as it has an impact on product characteristics such as the strength of the pinion. 5.2.2 Paper B: Analysis of process parameters during press quenching of bevel gear parts In transmission part manufacturing, heat treatment processes (for example case hardening) are commonly used to increase the strength and wear resistance of the produced parts (Lingamanaik & Chen, 2012; Arimoto, et al., 2011). A well known drawback of the case hardening processes is the more or less predictable geometric distortion (Holm, et al., 2010). These distortions have a broad range of possible causes with not fully known physical properties and mechanisms which make process control difficult (Funatani, 2011; Canale & Totten, 2005). In order to minimise distortion and control the quality, press quenching is a commonly used method in industry. Bagge and Lindberg (2012) has shown the potential to use DOE as a method for getting more knowledge about press quenching of bevel gears. The results enable definition of IPW tolerances satisfying a pre-defined acceptance level of defect parts. It would have been desirable to have a deep understanding of how material characteristics and phase transformations influence the result, but the aim of the paper was to find possible statistical relations between process control factors and their impact on dimensional properties. Figure 18 show the bevel gear crown wheel and press quenching set-up used for the DOE investigation.
65
34 | PERFORMED RESEARCH AND SYNTHESIS OF RESULTS
Back face
Ø240 Ø480
Cradle
Fexpander
Finner
Fouter
Quench oil flow
Figure 18: Investigated bevel gear crown wheel (top) and set-up in a press quenching machine (bottom).
The investigated properties are the centre-hole diameter (Ø240) and conical properties of the back face (dishing) according to Figure 19.
+
-
Figure 19: Crown wheel back-face dishing.
Despite the press quenching machine having a pressure setting for an expanding mandrel, aimed to influence Ø240, the effect is not obvious by studying historical production data. Data for Ø240 versus the corresponding expander pressure setting shown in Figure 20 does not correlate as expected.
PERFORMED RESEARCH AND SYNTHESIS OF RESULTS | 35
Ø240
Tolerance limits
Expander pressure
240,4
60
240,3 50 240,1
40
240,0
30 239,9 239,8
20
Pressure [bar]
Diameter [mm]
240,2
239,7 10 239,6 239,5
0
Figure 20: Historical production data for diameter 240 mm, compared to the corresponding expander pressure.
This is an example when the outcome of a manufacturing process is not simply deduced and understood by observing existing data. The DOE matrix included not only the expander pressure setting, but also six other factors supposed to influence the investigated responses: Ø240 and back face dishing. The multivariate method used for estimating regression models for both responses simultaneously was the partial least square method. The coefficients used in the models are shown in Figure 21. 0,04 0,04 0,03 0,02
0,01
0,01
[mm]
0,02
0,00
0,00
-0,01
-0,01
-0,02
-0,02 -0,03
-0,03
N=55 DF=48
R2=0,837 Q2=0,790
RSD=0,0236 Conf. lev.=0,95
R2=0,891 Q2=0,789
RSD=0,01809 Conf. lev.=0,95
Figure 21: Left: Coefficient plot and error bars (95%) for the response Ø240 Right: Coefficient plot and error bars (95%) for the response dishing.
Inner*Outer
Inner*Inner
Level
Mtrl_(M2)
Mtrl_(M1)
Cra_(C2)
Cra_(C1)
Inner N=55 DF=47
Outer
Level
Mtrl_(M2)
Mtrl_(M1)
Cra_(C2)
Cra_(C1)
Expan
Outer
-0,04
Inner
[mm]
0,03
36 | PERFORMED RESEARCH AND SYNTHESIS OF RESULTS
The regression models then facilitate estimations of the impact of individual factors and interactions of factors on the responses. The number of significant factors affecting each response (Ø240 and back face dishing) indicates the complexity of the process. The regression models have then been used to make predictions of the press quenching process outcome. Making process outcome predictions is a desirable feature from a process planning point of view. The predictions were made using the Monte Carlo simulation technique and facilitate the estimation of the number of parts out of specification, quantified as defect parts per million opportunities (DPMO). Bagge and Lindberg (2012) defined four scenarios which were examined regarding the estimated DPMO for both Ø240 and back face dishing. The scenarios were defined as combinations of two possible cradles (C1 and C2) and two different materials (M1 and M2). The predicted distributions and DPMO for each scenario are shown in Figure 22 and Figure 23.
C1 M1 DPMO=0
C1 M2 DPMO=0
C2 M1 DPMO=0
C2 M2 DPMO=0
10000
Counts
L. Tol.
U. Tol.
5000
0 239,95
240
240,05
240,1 Ø240 [mm]
240,15
240,2
Figure 22: Simulated distributions and DPMO for Ø240.
PERFORMED RESEARCH AND SYNTHESIS OF RESULTS | 37
C1 M1 0,04: DPMO=832 0,05: DPMO=26
C1 M2 0,04: DPMO=38 0,05: DPMO=0
C2 M1 0,04: DPMO=6 0,05: DPMO=0
C2 M2 0,04: DPMO=6 0,05: DPMO=0
20000
U. Tol.
Counts
L. Tol. 10000
0 -0,06 -0,05 -0,04 -0,03 -0,02 -0,01
0
0,01 0,02 0,03 0,04 0,05 0,06
Dishing [mm]
Figure 23: Simulated distributions and DPMO for dishing.
The DOE investigation and the consequent DPMO predictions are useful for the process planner because they give important information about the behaviour of the process and provide a good decision base. In this case, the possibility to tighten the tolerance for Ø240, block certain scenarios (C1 M1) or widen the tolerance for dishing can be useful when designing a well-balanced and competitive manufacturing process. To conclude Bagge & Lindberg (2012) : Design of experiments is a proper method capable of examining the complex process of press quenching. Regression models can in combination with Monte Carlo simulations be used to estimate the outcome of different press quenching scenarios. The simulation results make it possible to perform an IPW tolerance analysis by estimating DPMO for each scenario. Design of experiments is a method suitable to support process planning regarding process knowledge and tolerancing.
38 | PERFORMED RESEARCH AND SYNTHESIS OF RESULTS
5.2.3 Multivariate data analysis In many cases data from for example machine tool capability tests, cutting tool tests or measurements from running production are available after years of production. This data should be a good input to the process planner for improving an established process plan or bringing a new variant of a part into production. Depending on the complexity of the process to be evaluated, the diversity of parameters taken into account and data quality, data has to be handled in a more or less sophisticated way. For example, the relation between cutting tool wear, resulting in a gradually shifting dimension, and the produced number of parts may be good enough to predict process behaviour. However, if the result depends significantly on a diversity of factors such as the number of produced parts, material batch characteristics, surface roughness from an earlier operation, temperature in the workshop, etc., multivariate analysis is a more appropriate method. This approach, where production data is used as a base for finding the behaviour of a complex process, is used within the FFI project Realistic Verification (ReVer). The aim of the project is to arrange and analyse available production data in a way that geometrical distortions of a bevel gear crown wheel can be derived. The production data includes many manufacturing process parameter settings, material properties and information about the origin of the raw material. Unfortunately, it has shown to be a challenge to make the required multivariate data analysis and get reliable results. The reason is the great number of parameters having a potential impact on the geometrical distortion of the bevel gear crown wheel, and finding out which of them, and how they influence distortion, has been a topic for research for many years. Figure 24 shows a commonly recognized diagram where a multiplicity of parameters is identified, all with a potential to have an effect on the distortion.
PERFORMED RESEARCH AND SYNTHESIS OF RESULTS | 39
Cold forming Machining
Steel Composition Hardenability Grain size Rolling Refining Melt lot
Methods Lot Residual stress Shape Deform rate Equipment
Die set Machine Weight Fiber flow Core hardness Forge rate Surface hardness Post cooling Case depth Heating Shape Temperature Mass balance
Part design
Forging
Quenching
Press
Machinability Residual stress Cut volume Tool sharpness Cut condition Machine
Furnace Temperature Quenchant Agitation Pressure Tank design
Timing Cooling Temperature Duration Furnace Thermal cycle
Timing C-potenial Temperature Furnace Tray & Jig Atmosphere
Force Tool design Temperature Strategy Time
Distortion
Annealing
Carburizing
Furnace Cooling Duration Temperature
Pre heat treatment
Figure 24: Potential reasons for distortions. Based on Funatani, IMST Institute IDE2011, Bremen.
A large amount of production data has been sampled and organised for analysis in the ReVer project. The great number of potentially significant parameters in combination with uncertainty about the validity of some data makes an evaluation difficult. Despite these circumstances, some significant correlations between factors and responses have been identified within the project. The experiences show that it is not so easy to derive relevant parameters from the available production data without having a good hypothesis as a start. As in DOE, the chance of getting good results when analysing complex processes much depends on the initial knowledge and thoroughly reviewed assumptions. The better the hypothesis, the better the chance to succeed with the multivariate analysis. As much of the process planning work is knowledge intensive, multivariate analysis has a potential to be a good long-term contributor that knowledge. However, it requires a systematic approach to store production data with enough quality to be analysed. It also requires good skills in multivariate data analysis methods, as instantiated by SIMCA 4. 5.3 Tolerance synthesis and analysis Tolerance chains, or stack-ups, are well known within the product design domain but not usually recognised in production engineering, though they are undoubtedly present in many manufacturing processes. The appearance of tolerance chains and streams of variation (Ceglarek, et 4
SIMCA is software for multivariate analysis developed by Umetrics. www.umetrics.com
40 | PERFORMED RESEARCH AND SYNTHESIS OF RESULTS
al., 2004) in machining processes are similar phenomena to those found in assemblies. Design of multi-step manufacturing processes can be compared to product design of an assembly containing several parts. The assembly must be suited to the intended application and satisfy the needs of the customer. This requires that each constituent part of the assembly has an appropriate design and that they all are assembled in the right way. Putting the parts together often requires a specified sequence. An important aspect of designing assemblies is tolerancing of the parts. The part tolerances must be specified in a way that they, when combined, assure the function of the assembly. This combination of part tolerances forms one or several tolerance chains. In contrast, designing a multi-step process, such as for machining, does not aim to specify how to make an assembly, but how to combine different manufacturing processes to produce one final part in a proper way. This combination of manufacturing processes does often result in tolerance chains that determine how well the design part specification can be met. One important difference between design part tolerances and process plan tolerances must be emphasised. The product designer appends requirements to each design part as e.g. dimensional tolerances, and sometimes to the complete assembly. The design part tolerances are mainly on functional requirements arising from the intended application of the part. From the process planner’s point of view; the IPW tolerances must guarantee satisfaction of these requirements, but also process related needs in the different stages of manufacture. These processrelated functional needs are important to carry out for each manufacturing step. For example, in multi-step manufacturing where the workpiece is moved between fixtures in different machine tools, or just re-positioned and clamped in the same fixture, one or more features act as locating surfaces. These surfaces have the in-process function of making accurate fixturing of the workpiece possible. Another example is when a measuring operation requires special surface characteristics other than those defined by the product designer. The process planner has to ensure the quality of the final part, not by copying design part tolerances, but by defining both prerequisites and allowed outcome for each process step as IPW tolerances. These IPW tolerances are sometimes, but not always, equal to the design part
PERFORMED RESEARCH AND SYNTHESIS OF RESULTS | 41
tolerances and must, when combined, satisfy the design part requirements. This subject has been dealt with in two of the appended publications. The main development of a method for tolerance chain design and analysis was done by Bagge, et al. (2013) when first introducing the dimension dependency chart (DDC). The DDC was then further developed to become the DDC2 and used as a workbench for evaluation of different tolerancing strategies (Bagge, et al., 2014). 5.3.1 Paper C: Tolerance chain design and analysis of in-process workpiece Bagge, et al. (2013) highlights the relations between the process plan5 and the product specification, but also between the process plan and the basic manufacturing operations of which the process plan is composed. The relations between these basic manufacturing operations, called operation elements (OE), the process plan and the designed product specification are shown in Figure 25. Designed product specification
Comparison
Dimensions and tolerances Process plan
Operation element behaviour Operation element Operation element Operation element Operation element Operation element
Figure 25: Relations between operation elements, process plan and designed product specification.
5
“Process plan” represents in this case only the process steps, their sequence, nominal dimensions and tolerances. To be compared to the general description of process plan given in chapter 2 at page 5.
42 | PERFORMED RESEARCH AND SYNTHESIS OF RESULTS
The following gives a more detailed explanation of what an OE is: Each process step in the process plan defines a task to be performed, for example turning of a surface, and how the result is to be evaluated. Even if this process step is realised by just one turning operation, the method of evaluation may entail that the shown result depends on more than one operation. Typically, this situation appears when a distance between two machined surfaces needs to be measured and evaluated. The distance is a result of the position of two surfaces created in not only the last, but also in a previous operation. The definition of the process step therefore includes two single operations, which are the “operation elements“. The process plan is developed by defining and combining suitable operation elements capable of producing the final part. A method for the development of a process plan in terms of dimensions and tolerances is described and facilitated by the dimension dependency chart (DDC). The DDC is a structure based on the tolerance charting technique where the design part tolerances, the IPW tolerances and the operation element variations are connected. A conceptual description of the DDC is illustrated in Figure 26 where information flows between both design specification and process plan (upper view), and between OE and process plan (lower view) are shown.
Design specification
Upper view
Process plan
Process plan Lower view
Operation element behaviour Figure 26: A conceptual illustration of information flow in the dimension dependency chart (DDC).
PERFORMED RESEARCH AND SYNTHESIS OF RESULTS | 43
The DDC shown in Figure 27 has the capability to include and gather both operation element behaviours and process step definitions and make a comparison between the allowed outcome of the proposed manufacturing process and the design specification. Details of the procedure to establish a DDC are given in the appended paper C.
Symbols: Design dimension Machined feature and extent of dimension
-
+
F
Finally machined feature
X
Machining location and datum surface
Comparison 1
2
3
4
X
5
F
X X
F
Dim Tol 50 2
P. step 1
Dim Tol 50 2,3
20
2
20
0,1
20
3
20
2,8
20
4
20
1
Upper view
Dim Tol 71 1
P. step 2
49
1
P. step 3
30
0,5
1
1 -1
P. step 4
20
1
F
X
P. step 5
21
1,2
1
X
F
P. step 6
20
0,1
-1 1 -1
X
F
1 1
Comparison 1
2
3
4
X
5 F
X X
F
X
F
Tol 1
TE 0,5
SE 0,3
RE 0,2
P. step 2
1
0,4
0
0,4
P. step 3
0,5
0,4
0
0,4
P. step 4
1
1
0,6
0,4
P. step 1
F
X
P. step 5
1,2
0,5
0,3
0,2
X
F
P. step 6
0,1
0,2
0,1
0,1
Element
SE 0,3
RE 0,2
Element
0,3
0,2
Element
0,3
0,2
Element
0,3
0,2
X
F
X X
F
X
F
F
X
Element
0,3
0,2
X
F
Element
0,1
0,1
1
1
1
Lower view
1
-1 -1 1
Figure 27: DDC – Dimension dependency chart.
1 1
44 | PERFORMED RESEARCH AND SYNTHESIS OF RESULTS
The concluding remarks from Bagge, et al. (2013) are: The paper introduces the DDC for tolerance chain design and analysis of an IPW. The methodology is a novel approach based on the tolerance chart technique to define, include and connect operation element behaviour to the process plan and evaluate the expected outcome of the process chain. The DDC facilitates quantification and integration of smart process planning practice in the process plan and enables analytical calculation and evaluation of the results. 5.3.2 Paper D: Process chain based workpiece variation simulation for performance utilisation analysis As mentioned, the DDC was first introduced by Bagge et al. (2013) as a methodology for process planners to design and evaluate in-process tolerance chains. Figure 28 shows the second edition of the dimension dependency chart (DDC2) which is a result from elaborating the DDC methodology to facilitate the use and analysis of more substantial and detailed input data (Bagge, et al., 2014).
FD = Final dimension IPW = In-process workpiece PS = Process step OE = Operation element
Illustration of part geometry Comparison
Representation of product dimensions
Mapping matrix
Product dimensions & tolerances
PS
PS
PS
view of IPW variation
Representation of IPW dimensions
IPW dimensions & tolerances
OE
OE
view of IPW variation
Representation of IPW dimensions
FD Process chain dim. & tol.
OE IPW dimensions
Figure 28: DDC2 – Dimension dependency chart 2.
Mapping matrix
PERFORMED RESEARCH AND SYNTHESIS OF RESULTS | 45
The IPW tolerances for each PS are set by the process planner to limit the permissible IPW shape deviations. In addition to the IPW tolerances, it must be possible to satisfy the specification of the final part with the manufacturing process defined in the process plan. The process planner must evaluate each PS and the complete process chain regarding capabilities to fulfil all requirements. As discussed in section 2.2, the calculation of PCI is a commonly used method to indicate the abilities of a manufacturing process to produce within tolerance. Despite the statistical condition of normally distributed process outcome to estimate the standard deviation and calculate for example Cp, judgements based on PCI appears to be applied even when that condition is not met. Many manufacturing processes do show trends caused by tool wear, temperature changes, etc. and are often dependent with correlated outcomes implying non-normally distributed data. The first objective in the paper was to evaluate PCI as an indicator in analysing IPW tolerance chains, a task in planning of multi-step manufacturing processes. The second objective was to examine how the characterisation of causes of deviation in the process chain and its behaviour, affects the performance utilisation of the defined process. The research was performed by developing and using the DDC2 as a workbench for analysis of the process plan and evaluating how various strategies for tolerancing and use of process data result in different utilisation of the manufacturing equipment performance. As in paper C, the subject for the process plan was a shaft, here referred to as the “Xshaft”, shown to the left in Figure 29. The Xshaft is a fictitious part but it has distinctive features found in typical transmission parts like the gear shaft to the right in Figure 29.
20 ± x 20±x
20±x
50± x 1:1 Xshaft 02014
Figure 29: Xshaft (left) and gear shaft (right).
46 | PERFORMED RESEARCH AND SYNTHESIS OF RESULTS
The Xshaft is proposed to be machined in three set-ups. The contribution from each OE to the PSs defined in the process plan continues to the final part dimensions as illustrated in Figure 30.
Set-up 1 Set-up 2
Operation elements
Process steps
Final dimensions
OE1 OE2 OE3 OE4 OE5 OE6
PS1 PS2 PS3 PS4 PS5 PS6
FD1 FD2 FD3
FD4
Set-up 3
Figure 30: Manufacturing process for the Xshaft.
The focus for the research was to evaluate what can be achieved from the available manufacturing facilities when used as defined in the process plan. The manufacturing facilities have been assigned with behaviour characteristics, like tool wear and random variations, to represent industrial conditions. Each operation is defined as an OE and the ability to create a feature on the Xshaft depends on the assigned behaviour of that OE. The aim of the investigations is to find out the tightest tolerances on the FDs that the proposed manufacturing process chain can provide. The tighter the tolerance, the higher the performance of the process, and better the possibility for the part designer to increase the performance of the product. The evaluation criterion was the potential process performance utilisation (PUR) when different strategies for tolerancing and use of process behaviour definitions are applied to the process plan. Five simulation scenarios were defined. Four of the scenarios represent different IPW tolerancing strategies in combination with different characterisation of process behaviour information. The fifth scenario is the reference where the “pure process behaviour” and its propagation through the process chain determines the process outcome,
PERFORMED RESEARCH AND SYNTHESIS OF RESULTS | 47
without any limitations such as IPW tolerances and simplified representation of process behaviours. The reference scenario was defined to have 100% process PUR. All other scenarios showed to have lower or much lower PUR for one or more of the evaluated FDs. As this paper considers manufacturing processes showing not only random but also systematic variations, the significance of these systematic variations (expressed as SE) is of interest. What outcome will a virtual workshop with more pronounced SE than RE show compared to a workshop where SE and RE are similar? To show the impact of different workshop characteristics two examples were defined: virtual workshop 1 (VW1) and 2 (VW2). VW1 has SEs and REs with the same proportions; SE/RE quotient is approximately 1. VW2 has SEs double the REs and consequently the quotient is 2. The simulations were performed using Vensim DSS software (by Ventana Systems, Inc.) together with the DDC2. In total, 100 000 Xshafts were produced virtually for each scenario to avoid statistical uncertainties related to random factors. Figure 31 shows input and output data provided by Vensim for the simulation model based on the DDC2.
Dimension FD1
Dimension PS1
T1 T4
T3
71,04
50,04
Symbols:
T1
T2
Design dimension
T1
Machined feature and extent of dimension
-
F
+
Finally machined feature
X
Machining location and datum surface
50,01
71,01 1
70,98
1
Part instance no.
2
3
4
5
1 2 3 4
50 SEPS
REPS
F
X
-1 -1 -1 1 1 1 1 1 1
50 20
20
71
P. step 2
-49
P. step 3
-30
P. step 4
F
F
X
P. step 5
X
F
P. step 6
X
T1
X
T1
X
T2
X
F
1
Element Element Element
71 22
-1
-21
-1
20
-1 1 -1
-1
RE OE 1
1 1
41
1
Element
51
-1
F
X
Element
-21
-1
T4
X
F
Element
20
1
F
50
Part instance no.
-1
Normal direction
T3
F
-1
-20
Dim OE T1
1
20 DimPS
P. step 1
REOE
49,98
20
20 TolPS
X
DimFD
20
Design dim.
X
TolFD
Design dim. Design dim.
F
TolDP
50
TEPS X
SEOE
DimDP Design dim.
Dimension OE 1
0,03
0
50
71,04
Tool wear T1-T4 -0,03
0,06
71,01
Part instance no. 0,04
70,98
0,02
1
Part instance no.
50 0
1
Part instance no.
50
Figure 31: Vensim simulations applied on the DDC2.
48 | PERFORMED RESEARCH AND SYNTHESIS OF RESULTS
Performance utilisation ratio (PUR) Virtual Workshop 1 FD2
Virtual Workshop 2
FD4
FD3 FD1
FD2
Performance utilisation ratio
FD1
PS: Cp
Performance utilisation ratio
100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0%
100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0%
PS: TE
OE: TE
OE: SE&RE Reference
PS: Cp
PS: TE
FD3
OE: TE
OE: SE&RE Reference
FD4
100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0%
100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0%
PS: Cp
PS: TE
OE: TE
Scenario
OE: SE&RE Reference
PS: Cp
PS: TE
OE: TE
OE: SE&RE Reference
Scenario
Figure 32: Simulation results - Performance utilisation for FD1-4.
Figure 32 shows results from the simulations, in total 10 for each FD. It is important to make clear that VW1 and VW2 represent two different conditions to demonstrate how the workshop characteristics affect the result in general; they do not compete with each other. The aim is to do a sensitivity analysis regarding the impact on PUR related to the relation between SE and RE for the involved processes. A complete interpretation of the results is given in the appended paper. From the results the conclusions are: Using Cp in the IPW tolerancing strategy decreases the PUR, especially for high Cp values and long process chains. In the presented case study the use of TE instead of Cp=1,33 gives a higher PUR. Separating SE and RE is useful if there are coupled, or semi-coupled, OEs. SE for the PS will in these cases be eliminated or reduced. Simplified definitions of process behaviour result in an underestimation of the process performance. The longer the process
PERFORMED RESEARCH AND SYNTHESIS OF RESULTS | 49
chains, the bigger the underestimation. Detailed definitions and knowledge about process behaviour are vital for high PUR. PUR is a potential element of a key performance index for process planning. Additional comments on the DDC2 General methods for tolerance charting often include abilities to make stock removal calculations for cutting operations. Blank dimensions and tolerances together with stock removal calculations have also been a subject for integration into the DDC2. The attempts have shown to be useful in practice, but at the time of writing, no research results have been published. 5.4 Synthesis of results The triangular framework for process plan development (Figure 12) is synthesised as a result of the examined process planning methods described in sections 5.1-5.3. 5.4.1 Information flows Even though the process planning methods have been introduced and observed as separate issues, each one requires more or less information from the neighbouring subjects inside the triangle. There is also an important interaction and information exchange with the external collaboration parties “Part design”, “Workshop facilities” and “Manufacturing Technology”, adopted from Figure 13. If the available workshop facilities do not fulfil the needs, a development or investment in new manufacturing technologies must be considered. The specification of requirements on such technologies mainly derives from the proposed conceptual or schematic process plans. Information flows both inside and outside the triangle have been identified and are indicated in Figure 33 as arrows A-K.
50 | PERFORMED RESEARCH AND SYNTHESIS OF RESULTS
Manufacturing technology
Illustration of part geometry
Part design
Comparison
Product dimensions & tolerances
Representation of product dimensions
F Mapping matrix
G
PS
PS
PS
view of IPW variation
Representation of IPW dimensions
IPW dimensions & tolerances
OE
OE
view of IPW variation
Representation of IPW dimensions
Mapping matrix
J
OE IPW dimensions
D C
FD Process chain dim. & tol.
E
3 1
K
2
Finished part
C
b
A
B
B
a
Turning A’
b’ End-machining and B’ centre drilling
b’
a’
A
a’
Face driver
Retractable jaws
Parting line and residual flash
H
I
Workshop facilities
Figure 33: Triangular framework with information flows.
The information flows are provided by the capabilities of each subject in the triangular framework and aims to satisfy both internal and external needs during manufacturing process development. Table 3 specifies the information flows corresponding to the arrows A-K in Figure 33.
PERFORMED RESEARCH AND SYNTHESIS OF RESULTS | 51
Table 3 Information flows according to arrows A-K in Figure 33.
Internal information exchange:
A
Process sequence Operation elements Key-features
External information exchange:
F
Product specification: dimensions, tolerances and keyfeatures
B
Process behaviour IPW tolerances Product characteristics
G
Manufacturing prerequisites Part design suggestions Obtainable part tolerances Product characteristics Manufacturing knowledge (Manufacturing cost)
C
Process sequence Operation elements Measuring datums Machining datums IPW dimensions
H
Manufacturing resources Process characteristics Production data
D
IPW tolerances IPW dimensions Predicted outcome of process
I
Process plan Process settings
E
Process behaviour IPW tolerances
J
Manufacturing technology characteristics
K
Request for manufacturing technology
52 | PERFORMED RESEARCH AND SYNTHESIS OF RESULTS
5.4.2 Refining the PPP with information flows The PPP was first assigned in chapter 1 to describe the process planning domain as a process flow chart. To take the knowledge about process planning practice one step further, the content of the PPP can now be refined by adopting the identified information flows. These information flows indicated as A-K and the collaboration parties “Workshop facilities” and “Manufacturing technology” are brought into the PPP flow chart as shown in Figure 34 at page 53. However, the information flows cannot be unambiguously transferred to the PPP because it has a different structure to the triangular framework. The triangular framework represents methods connected to each other in terms of information flow interfaces. The PPP is a description of the process planning process where different activities, supported by methods, both require and are providers of information. This leads to some flows appearing in more than one place in the PPP and sometimes does not include all the elements of information defined in Table 3 at page 51. This is the case for information flow “I” which is found in two positions, but the information content for each position is clearly shown in the flow chart to be “Process settings” and “Process plan” respectively.
H
K
K
Figure 34: Information flows applied to the PPP.
Multivariate data analysis
Design of Experiments
Simulations
Process plan concept
Assignment directive
Initial process planning
F
Part design
H
A
Process settings
Process behaviour IPW tolerances
Product characteristics
I
G B E
Tolerancing
C
Workshop Manufacturing technology
Collaboration parties:
Schematic process planning
HJ
Defining process control strategy
IPW dimensions
IPW tolerances
FG
Part design
Process control strategy
Tolerance analysis
Schematic process plan
D Predicted process outcome
D
YES
Establish process plan
Process plan!
...
Measuring programs
NC-programs
Tool layout
Control documents
Work instructions
Examples of supporting processes to create:
Good enough?
NO
I
PERFORMED RESEARCH AND SYNTHESIS OF RESULTS | 53
54 | PERFORMED RESEARCH AND SYNTHESIS OF RESULTS
This page is empty
55
6 Discussion and conclusions 6.1 Coverage of the research results A number of methods to support process planning have been presented in Chapter 5. The coverage of all these methods on the PPP is summarised in Figure 35.
Part design
Part design
Schematic process planning
Schematic process plan
Assignment directive NO
Initial process planning
Process plan concept
Tolerancing
Defining process control strategy
In-process work piece tolerances
Tolerance analysis
Result
Good enough?
YES
Establish process plan
Process plan!
Process control strategy
Examples of supporting processes to create:
Simulations
Process behaviour
Work instructions Control documents Tool layout NC-programs
Design of Experiments
Process behaviour
Multivariate data analysis
Process behaviour
Known process behaviour
Measuring programs ...
Figure 35: Final coverage of presented methods on the original PPP.
There are two areas within the PPP indicated by dotted lines; multivariate data analysis and defining process control. These are included in the work but have been the focus of less research attention.
56 | DISCUSSION AND CONCLUSIONS
However, both areas are seen as important because they contribute to the design of competitive process plans. Multivariate analysis has a potential to contribute because it facilitates extraction of process behaviour from many different kinds of unstructured production data. However, it is a challenge to derive useful information if production circumstances have been changed and there is no possibility to trace the changes. A process control strategy is more or less mandatory and must be defined in harmony with the rest of the process plan, especially tolerances and measuring capabilities. The DDC2 methodology together with simulation software, like Vensim, can be used for evaluation of different control strategies. As shown in Paper B (Bagge, et al., 2014), using the same cutting edge for two conjugate surfaces reduces the need for process control without the risk of dimensional deviations. On the other hand, when this is not possible, there is still a potential to increase the utilisation of process performance by a well-defined process control strategy. This can be shown by assigning for example imperative tool change intervals and sequences for two or more cutting tools. It is also possible to integrate the effect of measuring and control strategies like statistical process control. The efforts put into process control can then be compared to the PUR. 6.2 Follow-up of research objective 1 The main result of the methodology studies is the triangular framework providing both qualitative and quantitative supporting methods for process planning. Figure 36 revives the illustration from Chapter 4 where the desirable characteristics of complete process planning methods are indicated as; descriptive, qualitative, analytic, quantitative and finally revealing.
DISCUSSION AND CONCLUSIONS | 57
As is As is
To be
Schematic process planning
Schematic process planning
Complete process planning
Descriptive, qualitative
Descriptive, qualitative
Descriptive, qualitative Analytic, quantitative Revealing
New research Capture of process behaviour
Capture of process behaviour
Perceptual, qualitative
Quantitative (Perceptual, qualitative)
Figure 36: Update of “Research and development of systematic process planning methods“.
The attained characteristic of each subject in the triangle does correspond to the conceptual representation of the research and development process leading to methods for “complete process planning”. Descriptive and qualitative: All methods in the framework have good potential to describe, explain and visualise the mechanisms and decisions behind a process plan. Both how single manufacturing process steps are defined and intended to work and how they interact when combined into process chains. All examined methods to capture process behaviour (simulation, DOE and multivariate data analysis) require some kind of computation software whose user interface considerably determines the descriptive qualities. Analytic and quantitative: The methods to capture process behaviour are all both quantitative and analytic, likewise the DDC2 for tolerance synthesis and analysis. Revealing: As both single manufacturing process steps and process chains can be analysed in terms of predicting the outcome, the methods have the capability to reveal the requested qualities of a process plan.
58 | DISCUSSION AND CONCLUSIONS
6.3 Follow-up of research objective 2 Interactions realised by the identified information flows must be emphasised as vital for all process planning work; either the work is done by systematic methods, automated or manual, or by an experienced process planner using intuitive, informal and undefined procedures. Many of the information flows can by intuition be defined by someone with process planning experience, but not surely with distinct reasons for why they exist and for what they are used. The information flows related to the framework in Figure 33 have their origins in verified demands and opportunities of the proposed process planning methods, which in turn support important process planning activities. These information flows, based on needs and opportunities, are used to refine and give more substance to the PPP, gaining better understanding of the process planning domain. 6.4 The relation between the PPP and PPAP Many manufacturing companies are subcontractors to original equipment manufacturers (OEM) that demand conformity in quality assurance methods and harmonized documentation of the results. An example of such a duty is the production part approval process (PPAP) (AIAG, 2006) often used to satisfy the international standard ISO/TS 16949. PPAP is also used by OEMs to assure quality and smooth deliveries from organisations within the company. The aim with PPAP is that the supplier must provide the evidence that the part will correspond to the design specification and possible to produce at a quoted production rate. Many of the required investigations to provide that information are included in the process planning work. The reason is simply that the process planning has the same goal as requested by the PPAP. How PPAP is related to the PPP is shown in Figure 37.
DISCUSSION AND CONCLUSIONS | 59
Approved PPAP
Production Part Approval Process (PPAP) Process flow diagram Control plan Initial process studies
Process Planning Process (PPP) NO
Process plan concept
Predicted process outcome
Good enough?
YES
Establish process plan
Process plan!
Significant production run
Result
Known process behaviour
Figure 37: The relation between PPP and PPAP.
Today, PPAP requires test-runs as “significant production run” or “full run test” to check the real outcome of the manufacturing process. By improving process planning methods to ensure reliable predictions of the process outcome, test-runs will probably be of less interest since they will show similar results to the predictions. This situation is also desirable for efficient process planning because the goodness of a proposed process plan can be estimated before it is finally established, without putting efforts into the creation of work instructions, control documents, NC-programs, etc. that must be changed. Another reason is that a test-run is performed during a limited period of time, without possibilities to test for example variations in material properties, workshop environment and idling machine-tools. Such sources for process variations are common in a workshop and may be important factors to analyse before start of production. 6.5 Process capability index As described in section 2.2, PCI is a commonly used measure of the capacity of a manufacturing process and is related to the balance between product and process performance. PCIs, including directives of required
60 | DISCUSSION AND CONCLUSIONS
levels, are more or less defined by both corporate standards and trade association standards like PPAP 4th ed. and APQP published by AIAG6. It is clear that for example Cm acts as a contributor of knowledge about the behaviour of one single machine tool and C p of a complete process in relation to the requirements defined by the process planner. But what is the effect of using that kind of aggregated and statistically constrained information as an evaluation criterion for manufacturing processes where the output not only depends on probability but on dependencies and predictable changes? One result in the appended publications (Bagge, et al., 2014) is that there is a potential to get better utilisation of the manufacturing process performance by sidestepping the established use of PCI, without increasing quality related deviations. As trends and non-normal distributed outputs can be recognised, not only for the IPW but for dimensions on the final part, the traditional use of PCI for FDs can be questioned also. This is also put in question by Wu, et al., (2009). The aim with PCI is to prevent poor quality by an indirect limit on the number of defects. The better the information is about process behaviour then better are the possibilities to predict the outcome and evaluate the risk of defects. Instead of traditional PCIs the use of estimated DPMO is more appropriate because it does not require normally distributed data. On the other hand, better the process knowledge, better the possibilities to control the process. This will reduce the impact from systematic errors and the random errors will dominate. This implies that the process output tends to be more normally distributed and the use of PCIs more appropriate. As discussed in section 2.5, the desirable balance between product performance and process performance can be expressed as some kind of PCI. A target value (not lower limit) for PCI, or better DPMO, should be a way for the process planner to design a well balanced manufacturing process. It is very important to find out the effect of using PCI in every process planning case, both regarding IPW tolerances and final part tolerances, to avoid low PUR.
6
Production Part Approval Process (PPAP) and Advanced Product Quality Planning (APQP) are published by the Automotive Industry Action Group (AIAG) at www.aiag.org.
DISCUSSION AND CONCLUSIONS | 61
6.6 Relevance of the chosen problems to solve During the work there have arisen some questions and hesitations related to the very broad range of activities that can be associated with process planning. One question is about the relevance of the identified problems to solve and the chosen objectives for the thesis. The first objective treated in this thesis has its origin in needs identified by myself but is not for what either process planners in industry or researchers usually ask. Process planners tend to be aware of time-consuming, regularised, activities that have to be done but do not require much process planning skill, like creation of work instructions and NC programming. This kind of activity can often be easily described and explained, but the decisionmaking behind it cannot. Probably this is why industry promotes supporting methods for these quite simple, but time-consuming, activities instead of the more abstract process planning core business. Academic research has in turn mainly been focusing on computer automated process planning, including optimisation, more than computer aided process planning in general terms. Even if many of the research results are promising, there is a problem to introduce them into an industrial process planning environment where few formal methods are defined, let alone standardised, and decision-making is a matter of personal experience and tacit knowledge. Process planners probably do not see how they should integrate the developed tools into their own work because of many undefined interfaces between activities and no possibility for plug-and-play. It is a risk that the effort of implementation of such tools is too high in relation to the perceived potential. The “perceived potential” is in turn a consequence of the process planner’s view of the overall scope of process planning and priorities, eventually underpinned by company key performance indicators (KPI). The full extent of a process planning scope depends on not only the part design, workshop facilities and new manufacturing technology but also the process planning work itself. Different decisions will result in more or less need for changes of the part design, rearrangements in the workshop or investments in new technology. As long as process planning decision-making is mostly heuristic and dependent on humans the decisions will be different in spite of the fact that the prerequisites were exactly the same. To understand the intention of the process planner at a
62 | DISCUSSION AND CONCLUSIONS
certain moment and make it possible to call a decision into question, there must be a proper representation of the basis for that decision. The records must be comprehensible, not only for a process planner but to provide further academic research. Even though most industrial process planners and academic researchers have not prioritised comprehensible process planning methods, the topic is still desirable and much work has to do be done within this area before CAPP can make its breakthrough. Ham (1988) did ask for more research facilitating a deeper understanding of process planning tasks 25 years ago and more currently, Xu et al. (2011) invites industry to play a more significant role in CAPP research. After all, the chosen problems to solve have their relevance both for academia and process planners in industry. 6.7 Scientific contribution As described in chapter 3.2, the unsuccessful implementation of CAPP in industry seems to be a consequence of a gap between industrial process planning practice and the academic research. Despite asking for both deeper understanding of process planning tasks (Ham, 1988) and a “scientifically rigorous foundation for planning methods” (ElMaraghy, 1993) for more than two decades, this gap is still un-bridged. The research presented in this thesis provides both the framework of process planning methods and the refined PPP as contributions to the establishment of a scientific foundation for process planning methods, aimed to fill the gap. The proposed framework of methods and the systematic representation of the process planning process, provided by the refined PPP, will also be an indirect support for future efforts in the development and implementation of CAPP systems. 6.8 General application of the research results The platform for the research has been manufacturing processes for gear transmission parts as discussed in section 2.6. Gear manufacturing processes have some special characteristics associated to the gear, but in general there are many similarities to manufacturing of other mechanical elements for engines, pumps, bearings, etc. Yet no manufacturing processes have been identified that would not be possible to develop by using the proposed methodologies. The PPP and
DISCUSSION AND CONCLUSIONS | 63
the triangular framework are supposed to be applicable in most mechanical precision manufacturing environments, especially where complex process chains appear. In parallel to the scientific contribution, the research results have a good potential to bring industrial process planning practice one step forward. Both a manufacturing process designed by more ingenious methods and the process planning work itself will gain better resource utilisation in terms of PUR and project lead time respectively. In addition, the potential to provide the product designer with relevant manufacturing information is of great interest. 6.9 Conclusions As process planning has a decisive impact on product quality and a considerably high influence on product cost, the possibilities to efficiently design and evaluate a process plan is of great interest. This thesis has treated the realisation of process plan design and evaluation by presenting several key contributors in the form of systematic methods. By gathering these methods into the triangular framework, it is possible to cover the process planning domain from initial process planning to prediction of the process outcome and finally establishing a process plan based on transparent and well scrutinised decisions. This corresponds very well to the first research objective. It has also shown that each subject contributes with necessary information for the rest of the framework, including the three collaborating parties; product design, workshop facilities and manufacturing technology. The second objective is satisfied by the refined PPP that provides a systematic representation of the process planning process. The refined PPP is a contribution to knowledge about the essence of manufacturing process planning. Three methods for capture of process behaviour have been discussed. One or more of these methods can be used to investigate process behaviour and bring valuable input data to the development of the process plan. The characteristics of each method are summarised in Table 4.
64 | DISCUSSION AND CONCLUSIONS
Table 4: Comparison of three methods to capture process behaviour. Simulations
Design of experiments
Multivariate analysis
When?
From predevelopment to mature process stages
From early development stages when test equipment is available
Established processes
How?
Assign FEM and/or analytical process models
Structured experiments, regression analysis
Gather production data, regression analysis
Resources
Software for appropriate modelling. Data base with for example material properties.
Process knowledge and test equipment
Process knowledge production data
Model
Process model Product model
Process model
Process model
Use
Simulation of process behaviour and prediction of process outcome. Prediction of secondary product properties.
Simulation of process behaviour and prediction of process outcome.
Simulation of process behaviour and prediction of process outcome.
Comment
Some processes are difficult to model. Requires deep knowledge about the mechanisms behind the process behaviour.
Requires some process knowledge and free time in test equipment
Requires large quantities of high quality production data
65
7 Future work 7.1 Economic aspects Despite cost being an important aspect when developing a manufacturing process it has only been discussed in general terms in this thesis. There is a need to combine the technical aspects of process planning with the economic aspect in a systematic way. The PPP and triangular framework adopt the important approach of covering the complete manufacturing process chain. They facilitate evaluation of how the aggregated technical contributions from the process steps involved affect the final part. This must be followed by an analysis of the aggregated economic contributions to the final manufacturing cost. One approach for this should be to combine the methods within the triangular framework with the technical-economic model described by Ståhl (2007), further developed and implemented by Jönsson, et al., (2008a; 2008b). If this integration can be done, it will be possible to define procedures for optimisation with an objective function on cost. Another approach is to use system dynamics cost simulations in a similar way to that Storck and Lindberg (2008) described for industrial production in a steel plant. 7.2 Representation of the PPP and information flows The representation of the triangular framework, the PPP and the information flows are made in a simple way in this thesis to enhance comprehensibility. However, a representation of activities, information flows, etc. following a standard such these in the family of IDEF7 modelling techniques, should facilitate easier transfer of the results into new research and development projects, especially in the context of computer modelling.
7
www.nist.gov
66 | FUTURE WORK
7.3 The model driven approach With some exceptions, CAD, CAM and CAPP solutions have not been used to support the proposed process planning methods. There are many well-established CAD and CAM systems on the market but their abilities to manage process planning related characteristics are limited. The benefit of CAM systems is the aid for creation and visualisation of NC programs for single machine tools, but not for linked multi-step manufacturing processes discussed in this thesis. A challenge for production research is to bridge the gap between CAD and CAM with systems capable to realize what has been presented in this thesis, most probably called CAPP. By adopting the research results in this thesis, it seems to be possible to develop a model driven approach, to establish a base for representation of process planning related information. Hedlind (2013) discusses the application of models to create, represent and use information of products, manufacturing processes and resources to support process planning. Further model based approaches with good potential are these for machining tolerance analysis discussed by Najad, et al. (2012). These approaches in combination with the proposed methods and structures for process planning presented in this thesis is suggested to be a subject for future research, heading for usable CAPP systems. 7.4 A new approach for P-FMEA Process failure mode and effects analysis (P-FMEA) is a step-by-step procedure to identify and evaluate risks of failure in a manufacturing process. The process is analysed regarding potential failures, the severity of the failures and the effect of the failures. As process planning is to define a process with a predictable outcome, the risk assessment should be a part of this work. The DDC2 facilitates representation of the process chain and has been used to analyse linked tolerances and process behaviour propagation. The DDC2 also has a potential to form a basis for process failure analysis by using the same structure as for tolerance analysis. The information needed for such assessment must therefore be integrated into the PPP and the DDC2 extended to support this kind of analysis.
67
8 References AIAG, 2006. Production part approval process (PPAP) 4th edition. s.l.:AIAG. Alting, L. & Zhang, H., 1989. Computer Aided Process Planning: the state-of-art survey. International Journal of Production Research, 27(4), pp. 553-585. Amaitik, S. M. & Kiliç, S. E., 2007. An intelligent process planning system for prismatic parts using STEP features. International Journal of Advanced Manufacturing Technology, Volume 31, pp. 978-993. Anderberg, S., 2012. Methods for improving performance of process planning for CNC machining - An approach based on surveys and analytical models. Göteborg: Chalmers University of Technology. Arimoto, K., Yoshii, T., Yokoto, H. & Narazaki, M., 2011. Experimental and simulation studies on distortions in steel plate specimens carburized on a single side and oil-quenched. In: H. Zoch & T. Lübben, eds. Proceedings of the 3rd International Conference on Distortion Engineering 2011. Bremen: SFB570, pp. 249-258. Azab, A. & ElMaraghy, H. A., 2007. Mathematical Modeling for Reconfigurable Process Planning. Annals of the CIRP, 56(1), pp. 467-472. Bagge, M., 2009. An approach for systematic process planning of gear transmission parts. Licentiate thesis, Stockholm, Sweden: Royal Institute of Technology. Bagge, M., Hedlind, M. & Lindberg, B., 2013. Tolerance chain design and analysis of in-process workpiece. In: A. Archenti & A. Maffei, eds. Proceedings of the International Conference on Advanced Manufacturing Engineering and Technologies. Stockholm: KTH Royal Institute of Technology, pp. 305-315. Bagge, M., Hedlind, M. & Lindberg, B., 2014. Process chain based workpiece variation simulation for performance utilization analysis. Submitted to Swedish Production Symposium 2014, Göteborg, Sweden, s.n. Bagge, M. & Lindberg, B., 2012. Analysis of process parameters during press quenching of bevel gear parts. In: M. Björkman, ed. Proceedings of The 5th International Swedish Production Symposium . Linköping: Produktionsakademien, pp. 251-259. Bose, U., 1999. A cooperative problem solving framework for computer-aided process planning. In: Proceedings of the 32nd Hawaii International Conference on System Sciences. 1 ed. Maui(USA): IEEE Computer Society.
68 | REFERENCES
Cai, N., Wang, L. & Feng, H. -Y., 2009. GA-based adaptive setup planning toward process planning and scheduling integration. International Journal of Production Research, May, 47(10), pp. 2745-2766. Canale, L. d. C. F. & Totten, G. E., 2005. Quenching Technology: A Selected Overview of the Current State-of-the-art. Materials Research, December, 8(No. 4), pp. 461-467. Ceglarek, D. et al., 2004. Time-Based Competition in Multistage Manufacturing: Stream-of-Variation Analysis (SOVA) Methodology—Review. International Journal of Flexible Manufacturing Systems, January, 16(1), pp. 11-44. Chang, T.-C., 2005. Tools for Manufacturing Process Planning. In: F. Kreith & D. Y. Goswami, eds. The CRC Handbook of Mechanical Engineering, Second Edition. Second edition ed. s.l.:CRC Press. Chen, D., Kjellberg, T., Svensson, R. & Sivard, G., 2011. Production Pilot for Co-operation in Factory Development. In: H. ElMaraghy, ed. Proceedings of the 4th International Conference on Changeable, Agile, Reconfigurable and Virtual production (CARV 2011). Montreal: CIRP, pp. 636-641. Chen, Y. D., Hui, A. & Du, R., 1995. A fuzzy expert system for the design of machining operations. International Journal of Machine Tools Manufacturing, 35(12), pp. 1605-1621. Denkena, B. et al., 2007. Knowledge Management in Process Planning. Annals of the CIRP, 56(1), pp. 175-180. DMMS, K., 2009. Produktionslotsen. [Online] Available at: www.produktionslotsen.se [Accessed 31 01 2014]. ElMaraghy, H. A., 1993. Evolution and future perspectives of CAPP. Annals of the CIRP, 42(2), pp. 739-751. ElMaraghy, H. & Nassehi, A., 2013. Computer-Aided Process Planning (CAPP). In: L. Laperrière & G. Reinhart, eds. CIRP Encyclopedia of Production Engineering. s.l.:Springer-Verlag. Fischer, B. R., 2011. Mechanical tolerance stackup and analysis. 2nd ed. Boca Raton: Taylor & Francis Group. Funatani, K., 2011. Distortion engineering with quench oil control for quality heat treatment. In: H. Zoch & T. Lübben, eds. Proceedings of the 3rd International Conference on Distortion Engineering 2011. Bremen: SFB570, pp. 101-108.
69
Groche, P. et al., 2012. Integration of manufacturing-induced properties in product design. CIRP Annals - Manufacturing Technology, Volume 61, pp. 163166. Gupta, D. P., Gopalakrishnan, B., Chaudhari, S. A. & Jalali, S., 2011. Development of an integrated model for process planning and parameter selection for machining processes. International Journal of Production Research, 1 November, 49(21), pp. 6301-6319. Halevi, G. & Weill, R. D., 1995. Principles of process planning - A logical approach. 1 ed. London: Chapman & Hall. Ham, I., 1988. Computer-Aided Process Planning: The Present and the Future. Annals of the CIRP, 37(5), pp. 591-601. Hamou, S., Cheikh, A., Linares, J. M. & Daho, A. C., 2006. A stochastic concept for the optimization of manufacturing tolerances in computer aided process plan simulation. International Journal of Computer Integrated Manufacturing, October-November, 19(7), p. 663 – 675. Hedlind, M., 2013. Model driven process planning for machining. 1 ed. Stockholm: KTH Royal Institute of Technology. Holm, T., Olsson, P. & Troell, E. eds., 2010. Stål och värmebehandling En handbok. Mölndal: Swerea IVF. Hua, G. & Xiaoliang, F., 2010. An intelligent approach of obtaining feasible machining processes and their selection priorities for features based on neural network. In: International Conference on Computational Intelligence and Software Engineering. Wuhan: IEEE, pp. 1 - 4. Inman, R. R. et al., 2013. Survey of recent advances on the interface between production system design and quality. IIE Transactions, Volume 45, p. 557–574. Jönsson, M., Andersson, C. & Ståhl, J. -E., 2008a. A General Economic Model for Manufacturing Cost Simulation. In: M. Mitsuishi, K. Ueda & F. Kimura, eds. Proceedings of the 41st CIRP Conference on Manufacturing Systems. Tokyo: Springer, pp. 33-38. Jönsson, M., Andersson, C. & Ståhl, J. -E., 2008b. Implementation of an Economic Model to Simulate Manufacturing Costs. In: M. Mitsuishi, K. Ueda & F. Kimura, eds. Proceedings of the 41st CIRP Conference on Manufacturing Systems. Tokyo: Springer, pp. 39-44. Kusiak, A., 1991. Process Planning: A Knowledge-Based and Optimization Perspective. IEEE Transactions on robotics and automation, June, 7(3), pp. 257266.
70 | REFERENCES
Lingamanaik, S. N. & Chen, B. K., 2012. The effects of carburising and quenching process on the formation of residual stresses in automotive gears. Computational Materials Science, 62(Sept 2012), pp. 99-104. Li, W. D. & Gao, L., 2010. Intelligent and cooperative manufacturing planning. International Journal of Manufacturing Research, 5(1), pp. 39-57. Li, W. D., Ong, S. K. & Nee, A. Y. C., 2002. Hybrid genetic algorithm and simulated annealing approach for the optimization of processplans for prismatic parts. International Journal of Production Research, 40(8), pp. 1899-1922. Li, W., Wang, L., Li, X. & Gao, L., 2011. Intelligent Optimisation for Integrated Process Planning and Scheduling. In: L. Wang, ed. Multi-objective Evolutionary Optimisation for Product Design and Manufacturing. London: Springer-Verlag, pp. 305-324. Lundgren, M., von Axelson, J., Hedlind, M. & Kjellberg, T., 2008. A systematic approach for product realization instructions in a model drivenenvironment: towards a three level instruction system development process. In: B. Lindberg & J. Stahre, eds. Proceedings of the Swedish Production Symposium 2008. Stockholm: Swedish Production Academy, pp. 319-324. Manoharan, T., Karuppan, S. & Govindharajalu, J., 2012. Selection of Machining Datum and Allocation of Tolerance through Tolerance Charting Technique. Chinese Journal of Mechanical Engineering, 25(4), pp. 697-705. Marri, H. B., Gunasekaran, A. & Grieve, R. J., 1998. Computer-Aided Process Planning: A State of Art. International Journal of Advanced Manufacturing Technology, Volume 14, pp. 261-268. Mayer, M. & Nusswald, M., 2001. Improving manufacturing costs and lead times with quality-oriented operating curves. Journal of Materials Processing Technology, Volume 119, pp. 83-89. Montgomery, D. C., 2009. Introduction to statistical quality control. 6 ed. New York: Wiley. Nejad, M. K., Vignat, F. & Villeneuve, F., 2012. Tolerance analysis in machining using the model of manufactured part (MMP) – comparison and evaluation of three different approaches. International Journal of Computer Integrated Manufacturing, 25(2), pp. 136-49. Qiao, L., Wang, X. -Y. & Wang, S. -C., 2000. A GA-based approach to machining operation sequencing for prismatic parts. International Journal of Production Research, 38(14), pp. 3283-3303. Rzevski, G., 1995. Artificial intelligence in engineering: Past, Present and Future. In: R. A. Adey, ed. Proceedings of the 10th International Conference on
71
Applications of Artificial Intelligence in Engineering (AIENG '95). Udine: s.n., pp. 1-14. Scallan, P., 2003. Process planning: the design / manufacture inteface. Oxford: Butterworth-Heinemann. Sterman, J. D., 2001. System Dynamics Modeling: Tools for learning in a complex world. California Management Review, Summer, 43(4), pp. 8-25. Storck, J. & Lindberg, B., 2008. A dynamic cost model for the effect of improved process flexibility in steel plants. In: M. Mitsuishi, K. Ueda & F. Kimura, eds. Proceedings of the 41st CIRP Conference on Manufacturing Systems. Tokyo: Springer, pp. 45-50. Ståhl, J. -E., 2007. Industriella Tillverkningssystem. Lund: Division of Production and Materials Engineering. Teti, R., Langella, A. & D'Addona, D., 1999. An intelligent computation approach to process planning in multiple-step cold forging. Annals of the CIRP, 48(1), pp. 175-178. Thimm, G., Britton, G. A. & Fok, S. C., 2004a. A graph theoretic approach linking design dimensioning and process planning Part 1: Designing to process planning. International Journal of Advanced Manufacturing Technology, Volume 24, pp. 261-271. Thimm, G., Britton, G. A. & Fok, S. C., 2004b. A graph theoretic approach linking design dimensioning and process planning Part 2: Design heuristics for rotational parts. International Journal of Advanced Manufacturing Technology, Volume 24, pp. 272-278. Werke, M., 2009. Simulation of manufacturing sequences for verification of product properties. Stockholm: Royal Institute of Technology. Werke, M., Bagge, M., Nicolescu, M. & Lindberg, B., 2014. Process modelling using upstream analysis of manufacturing sequences. The International Journal of Advanced Manufacturing Technology, Volume Under review April 2014. Wu, C.-C., Chen, Z. & Tang, G.-R., 1998. Component tolerance design for minimum quality loss and manufacturing cost. Computers in Industry, Volume 35, pp. 223-232. Wu, C. -W., Pearn, W. L. & Kotz, S., 2009. An over view of theory and practice on process capability indices for quality assurance. International Journal of Production Economics, Volume 117, pp. 338-359. Xu, H.-M., Yuan, M.-H. & Li, D.-B., 2009. A novel process planning schema based on process knowledge customization. International Journal of Advanced Manufacturing Technology, Volume 44, pp. 161-172.
72 | REFERENCES
Xu, X., Wang, L. & Newman, S. T., 2011. Computer-aided process planning - A critical review of recent developments and future trends. International Journal of Computer Integrated Manufacturing, January, 24(1), pp. 1-31. Zhang, L. L., Xu, Q. & Helo, P., 2012. A methodology integrating Petri nets and knowledge-based systems to support process family planning. International Journal of Production Research, 15 June, 50(12), p. 3192–3210.
